Identification and Molecular Characterization of an Efflux Pump Involved in Pseudomonas putida S12 Solvent Tolerance*

Bacteria able to grow in aqueous:organic two-phase systems have evolved resistance mechanisms to the toxic effects of solvents. One such mechanism is the active efflux of solvents from the cell, preserving the integrity of the cell interior.Pseudomonas putida S12 is resistant to a wide variety of normally detrimental solvents due to the action of such an efflux pump. The genes for this solvent efflux pump were cloned from P. putida S12 and their nucleotide sequence determined. The deduced amino acid sequences encoded by the three genes involved show a striking resemblance to proteins known to be involved in proton-dependent multidrug efflux systems. Transfer of the genes for the solvent efflux pump to solvent-sensitive P. putida strains results in the acquisition of solvent resistance. This opens up the possibilities of using the solvent efflux system to construct bacterial strains capable of performing biocatalytic transformations of insoluble substrates in two-phase aqueous:organic medium.

The microbial transformation of hydrocarbons is important not only in environmental applications such as soil remediation and waste stream purification, but also in biocatalytic applications for the production of specialty chemicals. The metabolic pathways by which many of these compounds are degraded in various bacteria have been elucidated and in many cases the genes coding for the enzymes involved have been cloned and sequenced (1). A major problem in applying hydrocarbon degrading bacteria to industrial processes is their susceptibility to the toxic effects of the very substrate that the organism is utilizing as a carbon source. This is often due to accumulation of the hydrophobic compound in bacterial membranes which can cause devastating effects on membrane structure (2,3). A second problem in the application of catabolic pathways in the synthesis of fine chemicals is that many of the desired substrates of enzymatic reactions are sparingly soluble in water and thus may not be fully bioavailable to microorganisms. The use of solvent tolerant bacteria allows the introduction of a nonpolar phase to the medium, dissolving the desired substrate, and increasing the exposure of the cell to the substrate.
Many different mechanisms have been described that contribute to solvent resistance ( Fig. 1) but despite these efforts no comprehensive overview is available to explain the physiological response of microorganisms to toxic organic solvents (for a recent review, see Weber and de Bont (4)). Our laboratories have been investigating the ability of Pseudomonas putida S12 to withstand toxic concentrations of toluene and other organic solvents (5). This organism has evolved at least two mechanisms to combat the accumulation of hydrophobic solvents in the membrane or the interior of the cell. One key observation was the detection of trans-rather than cis-unsaturated fatty acids in the membrane of the solvent-tolerant bacterium upon exposure to solvents (6,7). The conversion of cis-to transunsaturated fatty acids by a direct isomerization alters the packing of the phospholipids in the bacterial membrane. This results in a change in membrane fluidity, making the membrane less likely to allow solvents to partition into it, decreasing the detrimental effects on the membrane due to solvent partitioning, and thus increasing the solvent resistance of the cell (8). Recently it has been shown that a second mechanism of solvent resistance is possessed by P. putida S12. This is an energy-dependent active efflux system for solvents such as toluene (9) that may function in a fashion similar to that for multidrug efflux pumps found in many antibiotic-resistant microorganisms. Thus, P. putida S12 employs at least two mechanisms for active defense against the detrimental effects of solvents: one functioning to keep solvents out of the interior of the cell and a second functioning to prevent solvents from partitioning into the cell membrane.
In the present report we describe the construction of transposon mutants of P. putida S12 that have lost the solvent tolerant phenotype. This allowed the cloning of the genes responsible for a solvent efflux pump that is involved in the ability of P. putida S12 to withstand toxic concentrations of organic solvents. The nucleotide sequence of the genes involved was determined and their relationship to other bacterial efflux systems is discussed.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, Media, and Growth of Strains-P. putida S12 (10) is the wild-type strain capable of growth at supersaturated solvent concentrations (5) and is the object of the present investigation. P. putida JK1 is a solvent-sensitive mutant of P. putida S12 derived in the present work by transposon mutagenesis with TnMod-KmO. P. putida JJD1, also derived in the present work, contains a kanamycin gene/ColE1 origin cassette insertion in the genome of P. putida S12. The solvent-sensitive strain P. putida PPO200 is P. putida mt-2 cured of the TOL plasmid (11). The artificial transposable element TnMod-KmO 1 contains a kanamycin resistance gene and the ColE1 origin of replication between Tn5 inverted repeats. The Tn5 trans-posase gene and an origin of transfer for conjugation are present outside the inverted repeats. Escherichia coli JM109 (recA1 endA1 gyr A96 thi hsdR17 supE44 relA1 ⌬(lac-proAB) (FЈ traD36 proAB lacI q Z⌬M15) (12)) was utilized as the host strain for all recombinant plasmids. The cloning vector pUC19 (12), the pGEM series of cloning vectors (Promega, Madison, WI), and the broad host range vector pUCP22 (13) were used for the construction of subclones.
L broth (14) was used as complete medium. Minimal medium was prepared as described by Stanier et al. (15). Solid media contained 2% agar. Ampicillin (100 g/ml), kanamycin (50 g/ml), or gentamicin (25 g/ml) were added to the medium to maintain recombinant plasmids in E. coli. Gentamycin (25 g/ml) was used to maintain the pUCP22 subclones in P. putida and kanamycin (250 g/ml) was used in the selection of P. putida S12 transposon mutants. E. coli strains were routinely cultured at 37°C and P. putida strains were grown at 30°C. Growth of bacterial strains in the presence of various solvents (indicating solvent resistance) was determined essentially as described by Weber et al. (5). The experiments were conducted twice with duplicate samples each time.
Generation and Screening of TnMod-KmO Insertion Mutants-The conjugatable suicide transposon donor TnMod-KmO was introduced into P. putida S12 by triparental mating using pRK2013 as the mobilizing plasmid by established procedures (16,17). Kanamycin-resistant colonies were tested for the ability to grow on L agar plates in the presence of saturating vapor amounts of toluene. This was accomplished by placing the agar plates in a sealed glass dessicator along with a small beaker containing toluene. Growth was scored after 12 h at 30°C.
DNA Techniques-Total genomic DNA from P. putida strains was prepared by the CTAB procedure (18). Plasmid DNA was isolated by the alkaline-sodium dodecyl sulfate lysis method of Birnboim and Doly (19). DNA was digested with restriction enzymes and ligated with T4 ligase as recommended by the supplier (Life Technologies, Inc., Gaithersburg, MD). DNA restriction fragment and PCR 2 products were visualized by 0.7% or 1.0% agarose gel electrophoresis in 40 mM Tris, 20 mM acetate, 2 mM EDTA buffer. DNA from agarose gels was isolated using the method of Vogelstein and Gillespie (20). Plasmid DNA was introduced into either E. coli JM109 or P. putida JK1 cells by electroporation (21) using a Gene Pulser (Bio-Rad Laboratories).
All sequencing and PCR reactions were performed using a Gene Amp PCR System 9600 (Perkin-Elmer, Foster City, CA). Nucleotide sequencing reactions were performed with purified double strand plasmid DNA or PCR products using AmpliTaq FS DNA polymerase fluorescent dye terminator reactions (Perkin-Elmer) as recommended by the supplier. Sequencing products were detected using an Applied Biosystems 373A stretch automated DNA sequencer (Applied Biosystems Inc., Foster City, CA). Nucleotide sequence analysis was performed either with the Genetics Computer Group analysis package (22) or with the National Center for Biotechnology Information BLAST server (23). PCR reactions for amplifying the region of genomic P. putida S12 DNA contain-ing the insertion point for the transposon were performed using Taq DNA polymerase (Perkin-Elmer). The reaction mixture (100 l) was treated for 1 min at 94°C followed by 25 cycles of 1 min at 96°C, 1 min at 55°C, and 1 min at 72°C before finishing for 10 min at 72°C. Primers for this reaction were 5Ј-CGTTTGCAACCGGTGAG-3Ј and 5Ј-TATCGGACGCAAACG-3Ј corresponding to positions 3735 to 3752 and 4238 to 4253 of the nucleotide sequence, respectively.

RESULTS
Isolation of Solvent-sensitive Mutants-P. putida S12 was chosen for a molecular study of the basis of solvent resistance due to the extensive physiological studies that have been performed on the strain (4 -6, 9, 24 -26). The organism can grow in the presence of a wide variety of normally toxic solvents with log P OW values ranging from 2.3 to 3.5 (Table I). Initially, several solvent-sensitive transposon mutants were constructed using TnMod-KmO. P. putida S12 mutants which are no longer FIG. 1. Schematic representation of essential features of solvent-resistant bacteria. A, solvents diffuse to and preferentially partition to the cytoplasmic membrane where they cause disruptions in membrane functions by increasing membrane fluidity and affecting bilayer stability (3,4). B, to compensate for these effects, solvent-resistant bacteria modify the composition of the membrane either by isomerizing cis-into trans-unsaturated fatty acids from the membrane lipids (9,26) or by changing the headgroup composition (4,37). This compensation will only be partial and thus in a dynamic process, solvents have to be removed continuously from the membrane. C, removal to a certain extent may be by degradation (32). D, very recently, on the basis of whole cell experiments, evidence was obtained that an unprecedented energy-dependent export system for hydrophobic solvents is in operation (9). E, in combination with a retarded influx of solvents due to modifications at the outer membrane (31,44), the new efflux pump functions as the key factor in solvent tolerance.

TABLE I
Growth of P. putida S12 on L/acetate in the presence of various solvents Strain JK1 is a transposon mutant of the genes for a solvent efflux pump. pJD105 and pJD106 are clones containing the three genes for the solvent efflux pump (srpABC) cloned into the vector pUCP22. pJD105 has the genes in the same orientation as the lac promoter on the plasmid while pJD106 has the genes in the reverse orientation. Different solvents (1% final concentration) were separately added to identical subcultures in L medium with 60 mM acetate during the early exponential growth phase. Growth of the culture was measured at various times after solvent addition with no continued growth indicating solvent sensitivity. A plus indicates growth with OD Ͼ0.7 after 24 h while a minus indicates no growth with OD Ͻ0.1 after 120 h. Log P ow values were calculated according to the method of Rekker and de Kort (43) and is defined as the logarithm of the concentration of a solvent in octanol divided by the concentration of a solvent in water.

Solvent
Log P ow Growth on L/acetate and 1% solvent P. putida S12 resistant to solvents were detected by the inability of kanamycin-resistant exconjugants to grow on L medium in the presence of supersaturated vapor concentrations of toluene as described under "Experimental Procedures." Several toluenesensitive mutants were obtained and one of these, designated strain JK1, was chosen for further analysis. Besides toluene, JK1 is sensitive to a number of other solvents with log P OW values less than or equal to 3.5 (Table I), indicating that a single genetic trait is responsible for resistance to all of the solvents tested.
Cloning and Analysis of the Genes for Solvent Resistance-To characterize the genes for solvent resistance in P. putida S12 in more detail, the region of the genome containing the transposon insertion in mutant JK1 was cloned. This was aided by the fact that the TnMod-KmO transposable element utilized in the construction of the mutants contains an origin of replication derived from plasmid ColE1. Total genomic DNA from JK1 was cleaved with BamHI, ligated to form circular molecules, and electroporated into E. coli JM109 with selection for kanamycin resistance. Since BamHI does not cleave the transposon, the resulting clone, pJD101 (Fig. 2), must contain DNA from both sides of the transposon insertion. Approximately 11 kilobases of genomic DNA was cloned along with the transposon which contains the origin of replication and the kanamycin resistance gene (2 kb). The point at which the transposon is inserted is only 1 kb away from one end of the BamHI fragment cloned from strain JK1. This being the case, a second, overlapping 4-kb PstI fragment was cloned from strain JK1 (designated pJD102, Fig. 2).
The cloned genomic DNA was initially analyzed by determining the nucleotide sequence to either side of the transposon insertion point using primers specific for either end of the transposon. Screening for similar nucleotide sequences in the GenBank data base revealed a significant match with genes coding for multidrug resistance export pumps, consistent with the hypothesis that a solvent efflux pump is a key solvent resistance mechanism of P. putida S12. The complete nucleotide sequence of an operon into which the transposon had inserted was determined using a series of subclones and internal oligonucleotide primers. Both strands of a 6.5-kb SstI fragment were sequenced in their entirety. To accurately determine the nucleotide sequence at the transposon insertion point, a 0.5-kb DNA fragment was amplified by PCR from the genome of the wild-type strain P. putida S12 using oligonucleotide primers flanking the transposon insertion point in strain JK1. A cartoon of the nucleotide sequence obtained is shown in Fig.  2 to show the relationship of the open reading frames with the two clones pJD101 and pJD102 and the complete nucleotide sequence is shown in Fig. 3. The three open reading frames show significant homology to the three proteins that assemble to form proton-dependent multidrug resistance efflux pumps (27,28) and thus the genes were labeled srp for solvent resistance pump.
Complementation of P. putida JK1-To prove that the three open reading frames detected in the cloned fragment at the point of insertion of the TnMod-KmO transposon actually are responsible for solvent resistance, complementation experiments were performed. This required reconstruction of the operon since the clones obtained (pJD101 and pJD102) contain the transposon mutagenized DNA. A BglII-SstI kanamycin resistance cassette also containing the ColE1 origin of replication from TnMod-KmO was inserted at the BglII and SstI sites of a 6.7-kb ClaI fragment derived from pJD101. The resulting plasmid (pJD103, Fig. 2) was electroporated into P. putida S12 to construct a new mutation by site-specific reciprocal recombination. The resulting strain, JJD1, contains a kanamycin resistance gene adjacent to the srpABC genes in the genome. A 12-kb EcoRI genomic fragment containing the kanamycin/ ColE1 cassette was cloned from JJD1. This plasmid, designated pJD104 (Fig. 2), contains the intact srpABC genes. A 6.5-kb SstI fragment was cloned from pJD104 into the vector pUCP22 in both orientations with respect to the lac promoter. JK1 containing either of these two plasmids, designated pJD105 and pJD106 (Fig. 2), regained solvent resistance. JK1(pJD105), containing the srpABC genes in the same orientation as the lac promoter, regained resistance to all of the solvents that the original strain, P. putida S12, was resistant to (Table I). However, JK1(pJD106), containing the srpABC genes in the opposite orientation to the lac promoter, regained resistance to only two solvents, hexane and cyclohexane, with log P OW values near the border of the resistance phenotype. These results are consistent with the solvent resistance phenotype being dependent on the level of expression of the srpABC genes.
Transfer of the Solvent Resistance Phenotype-P. putida S12 displays multiple physiological responses to organic solvents (see Introduction). Intuitively, a solvent efflux pump would be the most important mechanism of solvent resistance since it would be involved in actively removing solvents from the cell. Experiments were therefore performed to determine whether the solvent efflux pump by itself is capable of imparting the solvent resistance phenotype on other P. putida strains. The two plasmids, pJD105 and pJD106, were electroporated into the normally solvent sensitive P. putida PPO200. The resulting recombinant strains are able to grow on rich medium in a toluene-saturated atmosphere, whereas the parent strain PPO200 with the vector pUCP22 could not (Fig. 4). PPO200-(pJD106) grew slightly slower than PPO200(pJD105), probably due to the fact that the srpABC genes are expressed from the lac promoter in pJD105. In liquid culture, PPO200(pUCP22) could withstand concentrations of toluene up to 2.8 mM while both PPO200(pJD105) and PPO200(pJD106) showed resistance to the toxic effects of toluene up to a concentration of 4.9 mM (Table II). Neither of the recombinant strains were resistant to a second phase of toluene (5.6 mM). These experiments indicate that the solvent resistance phenotype can be transferred to other bacterial strains and that the resistance can be enhanced by higher levels of gene expression. DISCUSSION In the past several years solvent-resistant microorganisms have been isolated directly from the environment (5, 29 -33) or through the process of mutation of a solvent-sensitive strain (34 -36). It is evident that these organisms must therefore have specific adaptation mechanisms that impart the solvent resistance. Physiological studies on microorganisms isolated from the environment that are naturally resistant to high levels of FIG. 3. Nucleotide sequence of an SstI fragment containing the srpABC genes. The deduced amino acid sequences of the encoded proteins are shown below the nucleotide sequence. Termination codons are indicated with an asterisk, putative ribosome-binding sites are underlined, a putative RNA polymerase-binding site is underlined, and the TnMod-KmO insertion point is marked with an arrow. The GenBank accession number for the srpABC sequence is AF029405.
organic solvents have revealed that many different factors may play a role (4,7). Naturally solvent-resistant bacteria have been shown to alter the composition of the cell membrane by increasing the ratio of trans-to cis-unsaturated fatty acids (26) or by changing the headgroup composition (37). This change in the cell membrane produces a physical barrier, preventing solvents from entering the cell by decreasing membrane fluidity. This would not entirely prevent solvents from entering the cell, only slow down their diffusion into the cell and increase the time needed to reach equilibrium with the external environment. An intuitively better method of solvent resistance would be to physically remove the solvent from the cell. One way of doing this would be to degrade the solvent but this would only be effective against low concentrations of solvents. Evidence was recently obtained that an energy-dependent export system for hydrophobic solvents functions to remove solvents from the interior of whole cells (9). This solvent efflux pump should be a key element in solvent resistance by naturally solvent-resistant bacteria. The genes for such a solvent efflux pump in P. putida S12 were identified via transposon mutagenesis to construct a solvent-sensitive strain. The genes were cloned and sequenced and their role in solvent resistance verified through complementation of the transposon mutation. The three genes involved were labeled srpABC for solvent resistance pump.
The deduced amino acid sequences of the proteins encoded by the srpABC genes have extensive homology with those for proton-dependent multidrug efflux systems of the resistance/ nodulation/cell division family (27,28). This "RND" family of efflux pumps is composed of three protein components that together span the inner and outer membranes of Gram-negative bacteria: an inner membrane transporter (SrpB analogues), an outer membrane channel (SrpC analogues), and a periplasmic linker protein (SrpA analogues). Members of this family have been shown to be involved in export of antibiotics, metals, and oligosaccharides involved in nodulation signaling. Based on the work presented here, this family can be broadened to include a new class of efflux pump, involved in export of solvents. Dendrograms showing the phylogenetic relationship of SrpA, SrpB, and SrpC to other proteins involved in multidrug resistance are shown in Fig. 5. The srpABC-encoded proteins show the most homology with those for the mexAB/oprMencoded multidrug resistance pump found in Pseudomonas aeruginosa (38,39). SrpA, SrpB, and SrpC are 57.8, 64.4, and 58.5% identical to MexA, MexB, and OprM, respectively. The three dendrograms show that the Srp and the Mex proteins fall into a distinct class, separate from but still closely related to the other members of the RND family of efflux pumps. The evolutionary relationship of the solvent resistance pump to multidrug resistance pumps is not surprising since they both FIG. 3-continued function to export hydrophobic molecules from the cell. It is logical that a solvent efflux pump would have evolved since it would enable microorganisms to survive in close proximity to oil or coal deposits (a rich source of carbon and energy) in the environment.
There have been several attempts to clone genes that are involved in solvent resistance leading in many cases to the identification of a protein that is somehow involved in making the cell resistant to a single solvent or to a group of related solvents. Most of these studies took place using E. coli that was forced under selective pressure to become solvent resistant. Mutations in a number of different genes can result in an increased tolerance to a particular chosen solvent, any one of which can allow the cell to grow in the presence of the solvent. Genes implicated in increased organic solvent tolerance of E. coli include the uncharacterized ostA (for organic solvent tolerance) (40), ahpC encoding alkylhydroperoxide reductase (35), robA encoding a global regulatory protein (41), and soxS encoding regulatory proteins controlling the superoxide response regulon (42). These genes enhance the survivability of the organism in the presence of a particular solvent but are not responsible for solvent resistance per se since they do not aid in understanding the true mechanism(s) of solvent resistance found naturally in environmental isolates. This article, however, represents the first example of cloning and characterization of genes for a major solvent resistance mechanism: a proton-dependent solvent efflux pump.
We have shown that the cloned genes can be transferred to another P. putida strain with the concomitant gain of solvent resistance by that organism. This has far reaching implications for industrial applications in the fine chemistry area. Existing and potential biocatalytic processes for compounds such as catechols, phenols, medium chain alcohols, and enantiopure epoxides are suboptimal because the products formed are very toxic to normal microorganisms. Product accumulation to a concentration which allows economic downstream processing is inherently prevented by the physical characteristics of these compounds. The ability, as demonstrated here, to take a normally solvent-sensitive strain and make it solvent resistant will greatly enhance the ability of a given strain to perform a desired biocatalytic reaction resulting in otherwise toxic products.