Abstract
MexJK requires OprM for erythromycin efflux but not for triclosan efflux. Deletion of 15 OprM family outer membrane proteins (OMPs) revealed that only the TolC homolog OpmH functions with MexJK for triclosan efflux. This is the first report of natural utilization of multiple OMPs by a given resistance nodulation cell division transporter/membrane fusion protein pair.
Efflux pumps of the resistance nodulation cell division family (RND) are associated with the intrinsic or acquired multidrug resistance phenotype of gram-negative bacteria (for recent reviews, see references 17 and 22). The RND pumps, as typified by AcrAB-TolC of Escherichia coli (8), form a tripartite complex spanning the entire cell envelope. High-resolution crystal structures of the outer membrane protein (OMP) component TolC (16), the RND membrane transporter AcrB (20, 26), and the Pseudomonas aeruginosa AcrA homolog MexA (1, 11) revealed a mechanism by which RND pumps can achieve this (7). Our previous studies showed that the MexJK efflux system required OprM for erythromycin efflux but not for triclosan efflux, and we reasoned that either MexJ and MexK were sufficient for triclosan efflux or MexJ and MexK interacted with an OMP other than OprM to form a functional triclosan efflux pump.
We previously determined that OprJ, OprM, and OprN did not function with MexJK to assemble an active triclosan efflux pump (4), although the respective OMP functioned with MexAB-OprM, MexCD-OprJ, and MexEF-OprN in triclosan efflux (2, 4). Because we suspected that an OMP of the 18-member OprM family might associate with MexJK, we decided to individually delete the genes encoding the remaining 15 members of this family in strain PAO327 constitutively expressing MexJK but not OprJ, OprM, and OprN. Although a comprehensive PAO1 transposon library was recently published which contains insertions in all OprM family genes (14), we decided to generate unmarked deletion mutants to avoid possible undesired effects resulting from truncated OMPs. The respective promoterless coding sequences were PCR amplified from PAO1 genomic DNA using gene-specific primers (unpublished data). Plasmid-borne deletions were generated and used to construct unmarked deletions in the PAO327 chromosome, utilizing published methods (12) (plasmids and strains used in this study are listed in Table 1; a comprehensive list of intermediate plasmids and strains is available from the authors). opmJ was deleted using a modified PCR-targeted method (5) in which the target gene is first cloned into a plasmid, followed by λRED-mediated recombination of a PCR-generated mutated copy of the gene (10). Because PCR amplification of the oprN gene proved futile for unknown reasons, an oprN::ISlacZ/hah-Tc transposon insertion from strain 8802 (14) was transferred to PAO327 using a previously described transformation procedure (9). Using these methods, we succeeded in deleting 14 of the 15 genes encoding the OprM family of OMPs and insertionally inactivating oprN. The resulting strains, except PAO375, contain well-defined unmarked deletion alleles and have between 21 and 395 codons deleted from the respective coding sequences. For complementation, the promoterless inserts were directionally cloned into pVLT35 (6), in which the DNA inserts are under the transcriptional control of the E. coli tac promoter and the lac repressor.
TABLE 1.
Strains and plasmids used in this study
| Strain or plasmid | Relevant genotype or propertiesa | Source or reference |
|---|---|---|
| PAO1 | wild type | 13 |
| PAO325 | Δ(mexAB-oprM)bnfxB Δ(mexCD-oprJ) Δ(mexJKL) Δ(mexXY) OpmH+ | 4 |
| PAO327 | Δ(mexAB-oprM) nfxB Δ(mexCD-oprJ) Δ(mexXY) mexL MexJ+K+ | 4 |
| PAO346 | PAO327 with ΔopmD302c | This study |
| PAO351 | PAO327 with ΔopmM81 | This study |
| PAO353 | PAO327 with ΔaprF153 | This study |
| PAO355 | PAO327 with ΔopmL133 | This study |
| PAO357 | PAO327 with ΔopmF123 | This study |
| PAO359 | PAO327 with ΔopmG151 | This study |
| PAO361 | PAO327 with ΔopmA21 | This study |
| PAO363 | PAO327 with ΔopmB282 | This study |
| PAO365 | PAO327 with ΔopmE209 | This study |
| PAO367 | PAO327 with ΔopmH362 | This study |
| PAO369 | PAO327 with ΔopmK53 | This study |
| PAO371 | PAO327 with ΔopmI184 | This study |
| PAO373 | PAO327 with ΔopmJ177 | This study |
| PAO375 | PAO327 with opmN::ISlacZ/hah-Tc | This study |
| PAO377 | PAO327 with ΔopmQ395 | This study |
| PAO386 | PAO325 with ΔopmH362 | This study |
| PAO397 | PAO386 with Δ(mexEF-oprN) | This study |
| 8802 | Tetr; PAO1 with opmN::ISlacZ/hah-Tc | 14 |
| KG4510 | Δ(nfxB-mexCD-oprJ) mexR ΔoprM Δ(mexXY) | 21 |
| pBSP II KS(−)/SK(−) | Apr; broad-host-range cloning vectors | 23 |
| pCR2.1 | Apr; TA cloning vector for PCR fragments | Novagen |
| pVLT35 | Spr/Smr; broad-host-range cloning vector | 6 |
| pPS1150 | Apr; pBSP II SK(−) with 6,945-bp NotI mexJKL fragment | 4 |
| pPS1266 | Apr; pCR2.1 with 1,673-bp opmH PCR fragment | This study |
| pPS1313 | Spr/Smr; pVLT35 with 1,664-bp PstI-HindIII opmH fragment from pPS1266 | This study |
| pPS1319 | Apr; pCR2.1 with 1,563-bp EcoRV-HindIII oprM PCR fragment | This study |
| pPS1369 | Apr; pCR2.1 SpeI(blunt)+XhoI with 4,369-bp SgfI(blunt)-XhoI mexJK fragment from pPS1150 | This study |
| pPS1370 | Apr; pBSP II KS(−) carrying a 4,394 bp BamHI-XbaI mexJK fragment from pPS1369 | This study |
| pPS1371 | Apr; pPS1370 BamHI(blunt)+HindIII with 1,530-bp EcoRV-HindIII oprM fragment from pPS1319 | This study |
| pPS1372 | Apr; pPS1370 BamHI(blunt)+HindIII with 1,659-bp PstI(blunt)-HindIII opmH fragment from pPS1266 | This study |
| pPS1373 | Apr; pPS1371 HindIII(blunt)+ClaI with 1,665-bp PstI(blunt)-ClaI opmH fragment from pPS1389 | This study |
| pPS1389 | Apr; pBSP II KS(−) with 1,664-bp PstI-HindIII opmH fragment from pPS1266 | This study |
| pPS1424 | Apr; pBSP II KS(−)XhoI(blunt)+HindIII with 1,659-bp PstI(blunt)-HindIII opmH fragment from pPS1266 | This study |
| pPS1425 | Apr; pBSP II KS(−)XhoI(blunt)+HindIII with 1,530-bp EcoRV-HindIII oprM fragment from pPS1319 | This study |
Abbreviations: Apr, ampicillin resistance; FRT, Flp recombinase target; Gmr, gentamycin resistance; Smr/Spr, streptomycin/spectinomycin resistance; Tetr, tetracycline resistance.
All deletions are unmarked but contain an FRT scar insertion; e.g., Δ(mexAB-oprM)::FRT.
Allele numbers indicate the number of codons deleted from the respective genes.
MICs for triclosan (KIC Chemicals, Armonk, NY) and erythromycin (Sigma, St. Louis, MO) were used as a measure of MexJK efflux pump activity in this panel of defined OMP mutants and their complemented derivatives. MICs were determined by the twofold broth microdilution technique (2). An agar incorporation method was used to determine triclosan MICs of >128 μg/ml (3). Deletion of only one gene, opmH, disrupted MexJK-mediated triclosan efflux, and overexpression of OpmH restored MexJK function to wild-type levels (Table 2). Deletion of opmH and overexpression of OpmH resulted in a marginal (twofold) decrease or increase, respectively, in the MIC of erythromycin, indicating that MexJK-OpmH may also be able to mediate some erythromycin efflux. Deletion of any of the other OprM family OMP genes and complementation with the corresponding genes expressed from the tac promoter did not have any significant effects on triclosan and erythromycin MICs, indicating that none of these OMPs can be utilized by MexJK to catalyze efflux of these antimicrobials. Reverse transcription-PCR analysis indicated that all genes were highly expressed in isopropyl-β-d-thiogalactopyranoside-induced cells, verifying that lack of complementation was not due to lack of expression (data not shown).
TABLE 2.
Contribution of oprM and opmH to erythromycin and triclosan efflux
| Strain/plasmid | OMP gene deletion/complementation | Efflux components expressed | MIC (μg/ml)a
|
|
|---|---|---|---|---|
| Tri | Ery | |||
| PAO327b | ΔoprJ ΔoprM | MexJK, OpmH | 128 | 16 |
| PAO367 | ΔoprJ ΔoprM ΔopmH362 | MexJK | 16 | 8 |
| PAO367/pVLT35c | ΔoprJ ΔoprM ΔopmH362 | MexJK | 16 | 8 |
| PAO367/pPS1313c | ΔoprJ ΔoprM | MexJK, OpmH | 128 | 16 |
| ΔopmH362/opmH+ | ||||
| PAO397/pBSP II-KS | ΔoprM ΔoprJ ΔoprN ΔopmH362 | None | 2 | 8 |
| PAO397/pPS1370d | ΔoprM ΔoprJ ΔoprN ΔopmH362 | MexJK | 2 | 8 |
| PAO397/pPS1371 | ΔoprM ΔoprJ ΔoprN ΔopmH362/oprM+ | MexJK, OprM | 2 | 32 |
| PAO397/pPS1372 | ΔoprM ΔoprJ ΔoprN ΔopmH362/opmH+ | MexJK, OpmH | >1,024 | 8 |
| PAO397/pPS1373 | ΔoprM ΔoprJ ΔoprN ΔopmH362/oprM+opmH+ | MexJK, OprM, OpmH | 512 | 64 |
| PAO397/pPS1424 | ΔoprM ΔoprJ ΔoprN ΔopmH362/opmH+ | OpmH | 8 | 8 |
| PAO397/pPS1425 | ΔoprM ΔoprJ ΔoprN ΔopmH362/oprM+ | OprM | 2 | 8 |
Erythromycin (Ery) MICs were determined using the broth microdilution method, and triclosan (Tri) MICs were determined using the agar incorporation method, except for PAO327 and the PAO367 series of strains, for which the microdilution method was used.
The genotype of PAO327 is Δ(mexAB-oprM) nfxB Δ(mexCD-oprJ) Δ(mexXY) mexL (MexJK constitutively expressed) and MexEF-OprN uninducible due to a mexT insertion (18).
In pVLT35 and pPS1313, opmH transcription is driven from the lacI repressor-controlled Ptac promoter; media used for MIC determinations with cells harboring these two plasmids were therefore supplemented with 1 mM isopropyl-β-d-thiogalactopyranoside (Gold Biotechnology, St. Louis, MO).
Hybrid mexJK (pPS1370), mexJK-oprM (pPS1371), mexJK-opmH (pPS1372), and mexJK-oprM-opmH (pPS1373) operons were assembled on pBSP II KS(−). pPS1371 contains oprM, 52 bp of its upstream region and 13 bp of its downstream region without transcriptional terminator. pPS1372 and pPS1373 contain opmH, 72 bp of its upstream region and 136 bp of its downstream region, including the transcriptional terminator. The recombinant plasmids were transformed into Δ(mexAB-oprM) Δ(mexCD-oprJ) Δ(mexEF-oprN) Δ(mexLJK) Δ(mexXY) ΔopmH362 strain PAO397, and efflux of the indicated antimicrobials was assessed by MIC determinations. On pPS1370-pPS1373, constitutive transcription of the cloned genes originates from the mexJK operon promoter (PJK). pPS1424 and pPS1425 are based on pVLT35, and transcription of cloned genes is driven from the lacI repressor-controlled Plac promoter; media used for MIC determinations with cells harboring these two plasmids were therefore supplemented with 1 mM IPTG.
If OpmH is indeed the OMP used by MexJK for triclosan efflux, then OpmH must either be constitutively expressed or be coexpressed with MexJK. Utilizing previously described procedures (21), rabbit anti-OpmH polyclonal antiserum was developed by immunization with three oligopeptides, VGDTRIAFDERPATVKRN (amino acids 77 to 94), LNQSEQSREGQRRQV (amino acids 341 to 355), and AEQLQSKPRQQY (amino acids 471 to 482), based on the deduced OpmH primary sequence (24). Membrane preparation and immunoblot analyses were performed as previously described (21). The anti-OpmH polyclonal antiserum did not cross-react with OprM, OprJ, OprN, OpmD, OpmE, and OpmB in membrane preparations of strains expressing these proteins constitutively but did react with native and histidine-tagged OpmH expressed in P. aeruginosa KG4510 (data not shown). Immunoblot analysis of total membrane proteins from PAO1 grown in LB medium (19) revealed growth phase-independent constitutive OpmH expression (Fig. 1). Similar observations were made when transcription of opmH was analyzed in LB-grown cells with an opmH′-lacZ transcriptional fusion and real-time PCR (data not shown). In both approaches, subinhibitory triclosan concentrations (8 μg/ml) did not significantly affect opmH expression, i.e., triclosan did not induce opmH transcription.
FIG. 1.
Constitutive production of OpmH in P. aeruginosa. Cells of PAO1 were grown in LB medium, and aliquots were removed at the indicated time points after subculture. Total membranes (5 μg of protein) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and subjected to immunoblot analysis using anti-OpmH polyclonal antiserum. Relative expression levels of OpmH were determined by quantifying band intensities using the Scion Image program (Scion Corporation, Frederick, MD). Each bar represents the mean ± standard deviation of the intensity levels obtained in four experiments. The average optical density (650 nm) of the cultures used for sample preparation is overlaid.
To further assess the roles of OprM and OpmH in MexJK-mediated erythromycin and triclosan efflux, we constructed hybrid operons containing mexJK alone, mexJK-oprM, mexJK-opmH, and mexJK-oprM-opmH, where constitutive transcription of all genes originated from the mexJK promoter. Plasmids containing these hybrid operons were transformed into strain PAO397, which carries deletions for five major efflux systems (including oprJ, oprM, and oprN) and the ΔopmH362 mutation. MICs for erythromycin and tetracycline were determined as a measure of efflux of the respective antimicrobials (Table 2). Expression of all genes was verified by reverse transcription-PCR (data not shown). It is evident that MexJK requires OprM for erythromycin efflux and OpmH for triclosan efflux, since in the absence of OprM and OpmH neither erythromycin nor triclosan efflux is observed. As expected, simultaneous expression of OprM and OpmH led to efflux of both antimicrobials, although triclosan efflux was somewhat lower when compared to efflux by the MexJK-OpmH pump. This is perhaps due to competition of OpmH and OprM for formation of a functional MexJK-OpmH triclosan efflux pump. Erythromycin efflux was marginally (twofold) higher in the strain expressing both OprM and OpmH when compared to efflux observed in the MexJK-OprM-expressing strain. Expression of OprM alone had no effect, indicating that the observed triclosan and erythromycin efflux was not due to the mere presence of this OMP alone or its interaction with yet another efflux system. In contrast, overexpression of OpmH resulted in a significant (fourfold) increase in the MIC for triclosan, indicating that OprM may be able to interact with yet another efflux system to mediate triclosan efflux. Of note also is that the triclosan MICs for PAO367 (16 μg/ml) and PAO397 (2 μg/ml) are significantly different, but we do not yet understand the cause for this difference.
The data presented here confirm that triclosan efflux in a strain expressing MexJK is mediated by a tripartite MexJK-OpmH efflux system. The constitutive expression and apparent absence of opmH cotranscribed efflux or secretion components confirm the previously published notion that the TolC homolog OpmH may play a more general role by interacting with more than one efflux system (15). However, the data presented here clearly indicate that these interactions may not be nonspecific, as previously suggested, but in some instances may rather depend on which substrate is being exported by the respective efflux system. How the OMP exerts its impact on substrate specificity of the pump remains unclear. One possibility is that the specific substrate export pathway, e.g., by opening of the exit pore, is formed only when all three proteins are properly assembled in the presence of the appropriate substrate. Another possibility is a substrate-induced assembly of the tripartite efflux pump, i.e., by recruitment of the appropriate OMP, similar to what has previously been observed with the HlyBD-TolC protein export system (25).
Acknowledgments
H.P.S. was supported by NIH grant AI051588, and N.G. was supported by grants from the Ministry of Education, Sports, Science and Technology and the Ministry of Health, Labor and Welfare, Japan.
REFERENCES
- 1.Akama, H., T. Matsuura, S. Kashiwagi, H. Yoneyama, S.-I. Narita, T. Tsukihara, A. Nakagawa, and T. Nakae. 2004. Crystal structure of the membrane fusion protein, MexA, of the multidrug transporter in Pseudomonas aeruginosa. J. Biol. Chem. 279:25939-25942. [DOI] [PubMed] [Google Scholar]
- 2.Chuanchuen, R., K. Beinlich, T. T. Hoang, A. Becher, R. R. Karkhoff-Schweizer, and H. P. Schweizer. 2001. Cross-resistance between triclosan and antibiotics in Pseudomonas aeruginosa is mediated by multidrug efflux pumps: exposure of a susceptible strain to triclosan selects nfxB mutants overexpressing MexCD-OprJ. Antimicrob. Agents Chemother. 45:428-432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Chuanchuen, R., R. R. Karkhoff-Schweizer, and H. P. Schweizer. 2003. High-level triclosan resistance in Pseudomonas aeruginosa is solely due to efflux. Am. J. Infect. Control 31:124-127. [DOI] [PubMed] [Google Scholar]
- 4.Chuanchuen, R., C. T. Narasaki, and H. P. Schweizer. 2002. The MexJK efflux pump of Pseudomonas aeruginosa requires OprM for antibiotic efflux but not for efflux of triclosan. J. Bacteriol. 184:5036-5044. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Datsenko, K. A., and B. L. Wanner. 2000. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. USA 97:6640-6645. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.De Lorenzo, V., L. Eltis, B. Kessler, and K. N. Timmis. 1993. Analysis of Pseudomonas gene products using lacIq/Ptrp-lac plasmids and transposons that confer conditional phenotypes. Gene 123:17-24. [DOI] [PubMed] [Google Scholar]
- 7.Eswaran, J., E. Koronakis, M. K. Higgins, C. Hughes, and V. Koronakis. 2004. Three′s company: component structures bring a closer view of tripartite drug efflux pumps. Curr. Opin. Struct. Biol. 14:741-747. [DOI] [PubMed] [Google Scholar]
- 8.Fralick, J. A. 1996. Evidence that TolC is required for functioning of the Mar/AcrAB efflux pump of Escherichia coli. J. Bacteriol. 178:5803-5805. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Gallagher, L. A., S. L. McKnight, M. S. Kuznetsova, E. C. Pesci, and C. Manoil. 2002. Functions required for extracellular quinolone signaling by Pseudomonas aeruginosa. J. Bacteriol. 184:6472-6480. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Gust, B., G. L. Challis, K. Fowler, T. Kieser, and K. F. Chater. 2003. PCR-targeted Streptomyces gene replacement identifies a protein domain needed for biosynthesis of the sesquiterpene soil odor geosmin. Proc. Natl. Acad. Sci. USA 100:1541-1546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Higgins, M. K., E. Bokma, E. Koronakis, C. Hughes, and V. Koronakis. 2004. Structure of the periplasmic component of a bacterial drug efflux pump. Proc. Natl. Acad. Sci. USA 101:9994-9999. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Hoang, T. T., R. R. Karkhoff-Schweizer, A. J. Kutchma, and H. P. Schweizer. 1998. A broad-host-range Flp-FRT recombination system for site-specific excision of chromosomally-located DNA sequences: application for isolation of unmarked Pseudomonas aeruginosa mutants. Gene 212:77-86. [DOI] [PubMed] [Google Scholar]
- 13.Holloway, B. W. 1955. Genetic recombination in Pseudomonas aeruginosa. J. Gen. Microbiol. 13:572-581. [DOI] [PubMed] [Google Scholar]
- 14.Jacobs, M. A., A. Alwood, I. Thaipisuttikul, D. H. Spencer, E. Haugen, S. Ernst, O. Will, R. Kaul, C. K. Raymond, R. Levy, L. Chun-Rong, D. Guenthner, D. Bovee, M. V. Olson, and C. Manoil. 2003. Comprehensive transposon mutant library of Pseudomonas aeruginosa. Proc. Natl. Acad. Sci. USA 100:14339-14344. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Jo, J., F. S. L. Brinkman, and R. E. W. Hancock. 2003. Aminoglycoside efflux in Pseudomonas aeruginosa: involvement of novel outer membrane proteins. Antimicrob. Agents Chemother. 47:1101-1111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Koronakis, V., A. Sharff, E. Koronakis, B. Luisi, and C. Hughes. 2000. Crystal structure of the bacterial membrane protein TolC central to multidrug efflux and protein export. Nature 405:914-919. [DOI] [PubMed] [Google Scholar]
- 17.Li, X. Z., and H. Nikaido. 2004. Efflux-mediated drug resistance in bacteria. Drugs 64:159-204. [DOI] [PubMed] [Google Scholar]
- 18.Maseda, H., K. Saito, A. Nakajima, and T. Nakae. 2000. Variation of the mexT gene, a regulator of the MexEF-OprN efflux pump expression in wild-type strains of Pseudomonas aeruginosa. FEMS Microbiol. Lett. 192:107-112. [DOI] [PubMed] [Google Scholar]
- 19.Miller, J. H. 1992. A short course in bacterial genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
- 20.Murakami, S., R. Nakashima, E. Yamashita, and A. Yamaguchi. 2002. Crystal structure of bacterial multidrug efflux transporter AcrB. Nature 419:587-593. [DOI] [PubMed] [Google Scholar]
- 21.Murata, T., N. Gotoh, and T. Nishino. 2002. Characterization of outer membrane efflux proteins OpmE, OpmD and OpmB of Pseudomonas aeruginosa: molecular cloning and development of specific antisera. FEMS Microbiol. Lett. 217:57-63. [DOI] [PubMed] [Google Scholar]
- 22.Poole, K. 2004. Efflux-mediated multiresistance in Gram-negative bacteria. Clin. Microbiol. Infect. 10:12-26. [DOI] [PubMed] [Google Scholar]
- 23.Schweizer, H. P. 2001. Vectors to express foreign genes and techniques to monitor gene expression in Pseudomonads. Curr. Opin. Biotechnol. 12:439-445. [DOI] [PubMed] [Google Scholar]
- 24.Stover, C. K., X.-Q. Pham, A. L. Erwin, S. D. Mizoguchi, P. Warrener, M. J. Hickey, F. S. L. Brinkman, W. O. Hufnagle, D. J. Kowalik, M. Lagrou, R. L. Garber, L. Goltry, E. Tolentino, S. Westbrock-Wadman, Y. Yuan, L. L. Brody, S. N. Coulter, K. R. Folger, A. Kas, K. Larbig, R. Lim, D. Spencer, G. K.-S. Wong, Z. Wu, I. T. Paulsen, J. Reizer, M. H. Saier, R. E. W. Hancock, S. Lory, and M. V. Olson. 2000. Complete genome sequence of Pseudomonas aeruginosa, an opportunistic pathogen. Nature 406:959-964. [DOI] [PubMed] [Google Scholar]
- 25.Thanabalu, T., E. Koronakis, C. Hughes, and V. Koronakis. 1998. Substrate-induced assembly of a contiguous channel for protein export from E. coli: reversible bridging of an inner-membrane translocase to an outer membrane exit pore. EMBO J. 17:6487-6496. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Yu, E. W., G. McDermott, H. I. Zgurskaya, H. Nikaido, and D. E. Koshland, Jr. 2003. Structural basis of multiple drug-binding capacity of the AcrB multidrug efflux pump. Science 300:976-980. [DOI] [PubMed] [Google Scholar]