Gene cloning and characteristics of the RND-type multidrug efflux pump MuxABC-OpmB possessing two RND components in Pseudomonas aeruginosa

Pseudomonas aeruginosa naturally exhibits a high level of resistance to a wide variety of agents (e.g. antibiotics, detergents and dyes) toxic to micro-organisms. In addition, this organism easily acquires higher resistance to anti-pseudomonal agents by multiple mechanisms (Dubois et al., 2008; Hocquet et al., 2008; Ikonomidis et al., 2008; Jalal et al., 2000; Lomovskaya et al., 1999; Pitout et al., 2008; Sobel et al., 2003). In the last few decades, emergence and spread of multidrug-resistant P. aeruginosa strains has been a great concern in clinical sites all over the world.

So far, several types of drug-resistant systems have been reported in P. aeruginosa. Among them, multidrug efflux pumps are now recognized to play an important role in the intrinsic and acquired resistance to structurally unrelated antimicrobial agents in P. aeruginosa.

Bacterial drug efflux pumps are generally classified into five families: the resistance nodulation cell division (RND) family, the major facilitator superfamily (MFS), the small multidrug resistance (SMR) family, the multidrug and toxic compound extrusion (MATE) family, and the ATP-binding cassette (ABC) family. The genome project for P. aeruginosa was completed in 2000, and the genome information predicted that there would be genes encoding 11 RND-type multidrug efflux pumps, 20 MFS-type multidrug efflux pumps, two SMR-type multidrug efflux pumps, one MATE-type multidrug efflux pump, and one ABC-type multidrug efflux pump (Stover et al., 2000). Among them, RND-type efflux pumps have been characterized in most detail.

In P. aeruginosa the RND-type efflux pump is thought to play an important role in the intrinsic resistance to antimicrobial agents from the result of gene disruptions (Morita et al., 2001b). In several clinically isolated mutants of P. aeruginosa, higher expression of RND-type efflux pumps was observed (Hirai et al., 1987; Rella & Haas, 1982; Wolter et al., 2004). Of the 11 RND efflux pumps identified in this organism, 10 pumps, MexAB-OprM (Poole et al., 1993), MexCD-OprJ (Poole et al., 1996), MexEF-OprN (Kohler et al., 1997), MexGHI-OpmD (Aendekerk et al., 2002; Sekiya et al., 2003), MexJK (Chuanchuen et al., 2002), MexMN (Mima et al., 2005), MexPQ-OpmE (Mima et al., 2005), MexVW (Li et al., 2003), MexXY (Mine et al., 1999; Westbrock-Wadman et al., 1999) and TriABC-OpmH (Mima et al., 2007), have been experimentally confirmed, and their properties have been reported. The RND-type efflux pump usually consists of three components to fulfil the function properly and includes an inner-membrane component (RND component), a periplasmic component (MFP component), and an outer-membrane component (OMP component) (Touze et al., 2004), and the three-dimensional structures of MexA, MexB and OprM have been reported (Akama et al., 2004a, b; Higgins et al., 2004; Sennhauser et al., 2009). However, regarding the most recently analysed RND efflux pump in P. aeruginosa, TriABC-OpmH, it has been reported that this system needs four components: an RND component, two MFP components and an OMP component (Mima et al., 2007).

PA2528-PA2527-PA2526-opmB encodes components for a possible RND-type efflux pump. This would be the last hitherto uncharacterized RND-type multidrug efflux pump in P. aeruginosa. The PA2528-PA2527-PA2526-OpmB system seems to possess two RND components. This type of RND-type efflux pump has been reported in Escherichia coli (MdtABCD) (Baranova & Nikaido, 2002; Nagakubo et al., 2002), but PA2528-PA2527-PA2526-OpmB in P. aeruginosa has not yet been characterized.

Here we report on the characterization of PA2528-PA2527-PA2526-OpmB (renamed as MuxA-MuxB-MuxC-OpmB) and on the isolation and properties of mutants that hyper-express MuxABC-OpmB.

Bacteria and growth.
Bacterial strains and plasmids used in this study are listed in Table 1. Cells were cultured at 37 °C under aerobic conditions. Bacterial strains were grown in L medium (1 % tryptone, 0.5 % yeast extract, 0.5 % NaCl, pH 7.0) (Lennox, 1955). Appropriate concentrations of antibiotics were added when required. PAI plates (2 % Bacto peptone, 1 % K₂SO₄, 0.03 % cetyltrimethylammonium bromide, 14.7 mM MgCl₂, 1 % glycerol, 1.5 % agar) were used to select P. aeruginosa cells. E. coli cells carrying pUCP20T (Schweizer, 1991; West et al., 1994), pEX100T (Hoang et al., 1998), or their derivatives were cultured in L medium with 100 µg ampicillin ml^–1. P. aeruginosa cells carrying pUCP20T (Schweizer, 1991; West et al., 1994) or its derivatives were cultured in L medium with 100 µg carbenicillin ml^–1.

Table 1. Bacterial strains and plasmids used in this study

Construction of plasmids.
Chromosomal DNA for a PCR template was prepared from cells of P. aeruginosa PAO1 (Chen & Kuo, 1993). The gene cloning for PA2528-PA2527-PA2526-opmB, designated muxABC-opmB, was performed in two steps (the primers used in this study are listed in Table 2). A PCR product containing the first half of muxABC-opmB was generated with Mux forward 1 and Mux reverse 1 primer pairs. The amplified fragment contains a putative promoter in the upstream region of muxA. Mux forward 2 and Mux reverse 2 primer pairs were used to clone the latter half of the operon. The amplified DNA fragment of the first half was digested with EcoRI and KpnI and ligated with pUCP20T, which was treated with EcoRI and KpnI. The resulting plasmid was designated pMUX92 (Fig. 1). The latter-half fragment of the muxABC-opmB operon was also cloned in pUCP20T [pMUX93 (Fig. 1)]. The pMUX92 plasmid was digested with KpnI and HindIII ligated to the KpnI–HindIII fragment (6.9 kbp) from pMUX93. The resulting plasmid carrying the whole region of muxABC-opmB was named pMUX2. Cells of E. coli KAM32 (Chen et al., 2002) were used for the plasmid construction. The plasmid constructs were transferred into P. aeruginosa cells for analysis.

Table 2. Primers used in this study

(23K): Fig. 1. Map of plasmids and location of ORFs. ORFs are indicated by thick black arrows. Regions of chromosomal DNA present in each recombinant plasmid are indicated by bold black lines. The white bar in pMUX82 indicates the mutant-type promoter region derived from the mutant P. aeruginosa PMX725. The positions of the primers are indicated by thin arrows above the ORFs. Plac with a small arrow indicates the lac promoter on the vector and its direction. F1, Mux forward 1; F2, Mux forward 2; F3, Mux forward 3; R1, Mux reverse 1; R2, Mux reverse 2.

For the construction of pMUX3, carrying muxABC-opmB but not the putative original promoter region, we used Mux forward 3 and Mux reverse 1 primer pairs for PCR. The product was digested with EcoRI and KpnI, and ligated with pUCP20T (pMUX94: Fig. 1). Plasmid pMUX3 was constructed by a similar procedure, using pMUX94 instead of pMUX92.

To investigate whether two RND components and an OMP component are necessary for the function of the efflux pump, we constructed pMUX4 (ΔmuxB), pMUX5 (ΔmuxC) and pMUX6 (ΔopmB). Plasmid pMUX2 was digested with SmaI and KpnI and blunted with T4 DNA polymerase (TaKaRa). Then the fragment was self-ligated, and the resulting plasmid was named pMUX4. Plasmid pMUX93 was digested with NotI and XhoI and blunted with T4 DNA polymerase. The fragment was self-ligated. Then the self-ligated plasmid was digested with KpnI and HindIII and inserted into the KpnI–HindIII site of pMUX92. The resulting plasmid was named pMUX5. The 0.32 kbp MefI fragment was deleted from pMUX2. The resulting plasmid was named pMUX6. The mRNA expression of the ORFs in the downstream region of the deleted gene was confirmed by RT-PCR in P. aeruginosa transformed with pMUX4 or pMUX5.

We also constructed pMUX8 and pMUX82 in order to investigate the effect of mutation at the promoter region of the muxABC-opmB operon. The strategy to construct pMUX8 and pMUX82 was common except for the PCR used to amplify the first half of muxABC-opmB. Primers Mux forward 1 and Mux reverse 1 were used to amplify the first half of muxABC-opmB in PMX7 or PMX725. Each PCR product was digested with EcoRI and KpnI and ligated into pUCP21T (Schweizer, 1991; West et al., 1994). Then the plasmid was digested with Eco52I to remove the Eco52I fragment (0.69 kbp). The plasmid deleted of the Eco52I fragment was digested with KpnI and XhoI, and the KpnI–XhoI fragment (1.2 kbp) from pMUX2 was ligated to the corresponding site. The plasmid digested with HindIII and KpnI was ligated with the HindIII–KpnI fragment (6.7 kbp) from pMUX2. We designated them as pMUX8 and pMUX82, possessing muxABC-opmB and the promoter region from PMX7 and PMX725, respectively.

Drug-susceptibility testing.
The MICs of various antimicrobial agents were determined in Mueller–Hinton broth (Difco) by the twofold dilution method according to CLSI recommendations (CLSI, 2006). Cells in the test medium (10⁵ cells ml^–1) were incubated at 37 °C for 24 h, and the growth was subsequently measured. Rokitamycin was a kind gift from Asahi Kasei Co. (Tokyo, Japan). The testing was repeated at least three times.

Isolation of novobiocin-resistant mutants.
Approximately 10⁹ cells of P. aeruginosa PMX7 were spread onto L plates containing 256 µg novobiocin ml^–1, which is eight times the MIC for P. aeruginosa PMX7. Five large colonies and approximately 100 small colonies appeared on the plate after 24 h incubation at 37 °C. We chose the five large colonies and one of the small colonies and analysed them.

RT-PCR.
RNA preparations and reverse transcriptional PCR were performed according to the manufacturer's protocols. Briefly, total bacterial RNA was isolated from cells grown to an OD₆₅₀ of 0.7 using the RNeasy Mini kit (Qiagen). Residual DNA was removed by the treatment with RNase-Free DNase (Promega). One nanogram of DNase-treated RNA was used as a template for one reaction using the Qiagen OneStep RT-PCR kit (Qiagen). Primer pairs to detect mRNA are listed in Table 2. Expression of the rpsL gene was used as an internal control. The products were separated by 3 % agarose X gel (Nippongene) electrophoresis and visualized with ethidium bromide.

Gene disruption in P. aeruginosa.
Plasmid pTAJ4, used to disrupt the mexVW genes, was constructed after several steps. First, pTAJ2 (Li et al., 2003) treated with SalI was ligated with SalI fragments (1.9 kbp) containing the gentamicin-resistance marker (aacC1) and sandwiched between two FRT (Flp recombinase target) sites from pPS858 (Hoang et al., 1998). The resulting plasmid pTAJ3 was digested with AflII and blunted; thereafter, the fragment was digested with SphI. The fragment produced (4.6 kbp) was ligated with pEX100T (Hoang et al., 1998) that had been digested with SmaI and SphI. The resultant plasmid was named pTAJ4, containing mexVW disrupted with the insert of aacC1 (9.9 kbp).

Plasmid pMUX2 was digested with SalI and the fragment (5.0 kbp) was self-ligated. Then the self-ligated plasmid was digested with EcoRI and HindIII. The EcoRI–HindIII fragment (0.8 kbp) was collected, and both ends were blunted with T4 DNA polymerase. The fragment was ligated with the SmaI site in pEX100T (pEMU1). The SalI fragment (1.9 kbp) from pPS858 was ligated into pEMU1 that had been treated with SalI. The resulting plasmid for muxABC-opmB disruption was named pEMU2.

We used the method of Hoang et al. (1998) to disrupt mexVW or muxABC-opmB. The plasmid for gene disruption was transferred from E. coli SM10 lacI^q (Schweizer, 1994) into P. aeruginosa by conjugation. The mexVW locus was disrupted using pTAJ4 in P. aeruginosa PMX6 (He et al., 2004), and the resulting strain was named P. aeruginosa PMX7. P. aeruginosa KN7 and KN725, which are muxABC-opmB gene disruptants, were constructed from P. aeruginosa PMX7 and PMX725, respectively. The genotypes of the resultant mutants were confirmed by PCR.

Drug susceptibility of P. aeruginosa possessing muxABC-opmB on a plasmid
Eleven RND-type efflux systems were predicted to be present on the P. aeruginosa PAO1 genome (). So far, 10 of these efflux systems have been characterized. Here we cloned the last hitherto uncharacterized genes for the putative RND-type efflux pump PA2528-PA2527-PA2526-opmB by PCR (Fig. 1). These genes appear to compose an operon that includes the opmB gene. OpmB has been previously reported to be an OMP component that cooperates with the MexAB, MexCD and MexXY multidrug efflux pumps (Murata et al., 2002). We renamed the PA2528-PA2527-PA2526 genes as muxA, muxB, muxC, respectively. The muxA gene seems to encode an MFP component. Both muxB and muxC seem to encode RND components. BLAST search showed that homologues for MuxABC were present in other bacteria. The most similar homologue among already characterized multidrug efflux pumps was MdtABC in E. coli (Baranova & Nikaido, 2002; Nagakubo et al., 2002). MuxA showed high similarity with MdtA, which is an MFP component (40 % identity and 78 % similarity). MuxB and MuxC showed significantly high similarity with the RND components MdtB and MdtC (65 % identity and 91 % similarity between MuxB and MdtB, 61 % identity and 88 % similarity between MuxC and MdtC).

To investigate the contribution of MuxABC-OpmB to resistance to several antimicrobial agents, pMUX2 carrying muxABC-opmB was introduced into cells of P. aeruginosa YM64 (Morita et al., 2001b). Since four major RND-type efflux pump genes are deleted in YM64, it is easy to assess differences in drug susceptibilities. P. aeruginosa YM64/pMUX2 showed increased MICs for aztreonam, novobiocin, tetracycline, erythromycin, kitasamycin and rokitamycin (Table 3). Increases in MICs were not observed with oleandomycin and clarithromycin, which are categorized as 14-membered macrolides (data not shown).

Table 3. OpmB is indispensable for the function of the MuxABC-OpmB efflux pump

A promoter-like sequence is present in the cloned fragment in pMUX2, and this putative promoter and the lac promoter on the vector are present in tandem in this plasmid. Since a potent repressor in P. aeruginosa cells might repress the gene expression even if the gene is cloned in a multicopy plasmid, we constructed pMUX3, carrying muxABC-opmB without the putative promoter. In this plasmid muxABC-opmB is present just downstream of the lac promoter. P. aeruginosa YM64 transformed with pMUX3 showed the same MIC values of several antibiotics as YM64/pMUX2 (data not shown).

Necessity for two RND components, MuxB and MuxC
Since MuxABC shared sequence similarity with MdtABC, an RND-type efflux pump of E. coli, the properties of these two systems might be similar. In E. coli, it has been reported that both of the two RND components, MdtB and MdtC, were essential for its full activity (Baranova & Nikaido, 2002; Nagakubo et al., 2002). To investigate the necessity for MuxB and MuxC, we constructed pMUX4 (ΔmuxB from pMUX2) and pMUX5 (ΔmuxC from pMUX2) (Fig. 1). The MICs of several drugs such as aztreonam and novobiocin in P. aeruginosa YM64 transformed with pMUX4 or pMUX5 were the same as those in the control cells, YM64/pUCP20T (data not shown). We should take into account a possible polar effect on the genes located downstream of the disrupted gene. Therefore we measured mRNA expression of the genes that are downstream of the deleted ORF. No change was observed in the expression of the genes adjacent to the deleted gene (data not shown). Therefore, we conclude that both of the two RND components are essential for the function of the MuxABC-OpmB pump.

We also constructed a plasmid with muxA deleted from pMUX3 to test the necessity of MuxA for the pump function. P. aeruginosa YM64 transformed with this plasmid did not show increased MICs of aztreonam and novobiocin (data not shown). Thus, we conclude that MuxA is also essential for the pump.

Necessity for OMP component OpmB
The RND-type efflux pump requires an OMP component for function (Aires et al., 1999; Fralick, 1996). The opmB gene encoding an OMP component is present in the downstream region of muxABC. We showed that introduction of muxABC-opmB into cells of P. aeruginosa resulted in elevated MICs of several antibiotics (Table 3). This suggested that OpmB could function with MuxABC. We investigated, therefore, whether OpmB is really essential for the function with MuxABC. We constructed a plasmid pMUX6 carrying an intact muxABC region and a disrupted opmB gene (Fig. 1). P. aeruginosa YM64 transformed with pMUX6 did not show increased MICs of antimicrobial agents tested compared with the control YM64/pUCP20T (Table 3). Thus, we conclude that OpmB is essential for the pump function of MuxABC. It should be pointed out that we observed somewhat decreased MICs of aztreonam, novobiocin, tetracycline, erythromycin, kitasamycin and rokitamycin with YM64/pMUX6 compared to the control (Table 3). This suggests that OpmB also functions with other multidrug efflux pump(s) that extrude these antibiotics. The possible reason is discussed later.

YM64 host cells are defective for oprM, which codes for OprM, a major OMP component for RND-type multidrug efflux pumps in P. aeruginosa. Since OpmB showed high similarity with OprM, we tested the possibility that MuxABC can use OprM instead of OpmB, by using YM44, an OprM active strain, as host. Contrary to our expectation, however, we observed no increase in the MICs of antimicrobial agents tested with YM44/pMUX6 (possessing muxABC and oprM) compared with YM44/pUCP20T (control possessing oprM) (Table 3).

These results support the view that OpmB cannot be replaced with OprM for the function of the MuxABC system.

Isolation of novobiocin-resistant mutants
It seemed possible to isolate mutants that show elevated expression of the muxABC-opmB operon or elevated function of the MuxABC-OpmB pump. For this purpose, we used P. aeruginosa PMX7 (ΔmexAB-oprM, ΔmexCD-oprJ, ΔmexEF-oprN, ΔmexXY, ΔmexHI-opmD, ΔmexVW, ΔpmpM). Since novobiocin is a good substrate for the MuxABC-OpmB pump (Table 3), we tried to isolate novobiocin-resistant mutants from PMX7. The MIC of novobiocin in all six mutants was elevated, and was eight times higher than that in the parental PMX7. Four of them showed elevated MIC of only novobiocin (data not shown). However, in two remaining mutants, the MICs of aztreonam, erythromycin, kitasamycin and rokitamycin were also elevated (Table 4). We named one of these multidrug-resistant mutants as PMX725. PMX725 grew better in the presence of tetracycline than PMX7 (data not shown) although the MIC of tetracycline in these two types of cells was the same (Table 4).

Table 4. Contribution of MuxABC-OpmB to the resistance to antimicrobial agents

Since novobiocin, aztreonam, erythromycin and tetracycline are substrates for the MuxABC-OpmB multidrug efflux pump, PMX725 is presumed to be a mutant with respect to this pump. In order to test whether muxABC-opmB is responsible for elevated resistance to these antibiotics in PMX725, we disrupted muxABC-opmB in the mutant. As expected, the construct KN725, a disruptant of muxABC-opmB derived from PMX725, showed reduced MICs of the antibiotics compared with the parental strain PMX725 (Table 4). Moreover, the result of the RT-PCR measurement revealed that the expression of muxA increased in PMX725 (Fig. 2). These results indicate that the increased resistance to novobiocin, aztreonam and erythromycin in the mutant PMX725 was caused by the increased expression of muxABC-opmB.

(29K): Fig. 2. mRNA expression of muxA in P. aeruginosa. Expression of muxA (upper) and rpsL (lower; internal control) was measured by the RT-PCR method. Lanes: 1, marker (pUC19 digested with Msp I); 2, PMX7; 3, PMX725. The PCR cycles were 30 for muxA and 24 for rpsL.

Interestingly, KN7, a deficient mutant of muxABC-opmB genes, was more susceptible to novobiocin, kitasamycin and rokitamycin (which are substrates for the MuxABC-OpmB pump) than the parental PMX7 (ΔmexAB-oprM, ΔmexCD-oprJ, ΔmexEF-oprN, ΔmexXY, ΔmexHI-opmD, ΔmexVW, ΔpmpM) (Table 4). In addition, an RT-PCR experiment revealed that muxA was expressed in the parental PMX7 (Fig. 2). These results also indicate that muxABC-opmB is expressed in the parent, and is responsible for resistances to these antibiotics in the parental PMX7.

Identification of the mutation in PMX725
In order to identify the mutation in P. aeruginosa PMX725 in which expression of the muxA gene was elevated (Fig. 2), we determined the DNA sequence of the putative promoter region of muxA in PMX725. A single nucleotide insertion was found adjacent to the putative –10 region for the muxABC-opmB operon in the PMX725 strain (Fig. 3).

(9K): Fig. 3. Nucleotide sequence of wild-type and mutant-type promoter region of muxABC-opmB. The upper row indicates the nucleotide sequence in wild-type PAO1, and the lower row indicates that in PMX725. The deduced promoter region is shown as –35 and –10, and the single nucleotide insertion (T) near the putative –10 region is indicated by a box.

To confirm whether the insertion of T adjacent to the possible –10 region of the putative promoter for muxA in PMX725 is responsible for increased expression of the muxABC-opmB operon, we cloned muxABC-opmB from PMX725 with the mutated promoter region (pMUX82). We compared the expression levels of the operon by RT-PCR. Expression of muxA, muxB, muxC and opmB in KN7/pMUX82 (mutant type) was clearly higher than that in KN7/pMUX8 (wild-type) (data not shown). Based on these results, we conclude that the insertion of T in the putative promoter region of muxABC-opmB enhanced expression of the genes in PMX725, and that the putative promoter would be the real promoter for the genes. In this study we cloned the PA2528-PA2527-PA2526-opmB genes encoding an RND-type multidrug efflux pump and renamed PA2528, PA2527, PA2526 as muxA, muxB and muxC, respectively. This pump system needs four components, two RND components (MuxB and MuxC), one MFP component (MuxA) and one OMP component (OpmB). MuxABC-OpmB is the only RND-type pump that requires two RND components in P. aeruginosa. We surveyed genome databases to find genes encoding two RND-component-like proteins in an operon, and found homologues of MuxB and MuxC in Ralstonia [e.g. R. solanacearum (Salanoubat et al., 2002), R. pickettii (NC_010678)], Chromobacterium [e.g. C. violaceum (Brazilian National Genome Project Consortium, 2003)], Pseudomonas [e.g. P. putida (YP_001267510), P. entomophila (Vodovar et al., 2006), P. fluorescens (Paulsen et al., 2005), P. syringae (Almeida et al., 2009)], Burkholderia [e.g. B. dolosa (YP_002098778), B. thailandensis (YP_443379), B. pseudomallei (YP_002106659), B. cenocepacia (YP_001764468)] among others. The function of these homologues, however, has not been investigated. Thus, it is unclear whether they are drug efflux pumps.

MuxABC shows high similarity with MdtABC from members of the Enterobacteriaceae (Baranova & Nikaido, 2002; Kube et al., 2008; Nagakubo et al., 2002; Nishino & Yamaguchi, 2001; Nishino et al., 2007; Parkhill et al., 2001a, b). It has been reported that MdtABC elevated the resistance levels against cholate, SDS and novobiocin in E. coli (Baranova & Nikaido, 2002; Nagakubo et al., 2002). We tried to measure the MICs of cholate and SDS in P. aeruginosa YM64/pUCP20T (control) and YM64/pMUX2 (carrying muxABC-opmB), but were unable to because the cells grew at the solubility ranges of cholate and SDS.

We detected the mRNA of muxABC-opmB in cells of wild-type P. aeruginosa PAO1 (data not shown). Therefore it seemed possible that the MuxABC-OpmB pump contributes to the intrinsic resistance of P. aeruginosa against novobiocin, and to the macrolides to some extent. However, we were unable to detect a change in the MIC of novobiocin when we disrupted the muxABC-opmB genes in P. aeruginosa PAO1 (data not shown). The reason why an MIC change was not detected in the mux-deleted cells could be that potent novobiocin efflux pumps such as MexAB-OprM and MexCD-OprJ are active enough in PAO1 (Masuda et al., 2000; Schweizer, 2003) to mask the defect of MuxABC-OpmB.

Our results indicated that the elevated MIC of novobiocin in mutant PMX725 was caused by a mutation in the promoter of the muxABC-opmB operon. A similar result was recently reported for triABC in P. aeruginosa (Mima et al., 2007). These authors reported that a single nucleotide substitution in the predicted promoter –35 sequence elevated the expression of triABC. It has also been reported in various bacteria that a mutation in a promoter region enhanced gene expression related to drug resistance and that the cells showed increased drug resistance (Huang et al., 2004; Kaatz et al., 1999; Siu et al., 2003). Therefore, it seems to be quite a common phenomenon that mutations in the promoter region elevate drug resistance levels in bacteria.

Lack of an OMP component of the RND-type pump confers a severe effect on the resistances to antibiotics in bacteria, and it is sometimes more serious for cells than the lack of an RND component for a drug efflux pump (Morita et al., 2001a; Sulavik et al., 2001). Therefore, we investigated whether the decrease in the MIC of novobiocin in KN7, the whole muxABC-opmB of which is disrupted, was mainly caused by the deletion of opmB alone. The MIC of novobiocin was measured in KN7 transformed with a plasmid carrying opmB but not muxABC. We observed no change in the MIC value by the introduction of opmB (data not shown). Thus, it is clear that not only OpmB, but also MuxABC, are necessary for the observed novobiocin resistance. It should be pointed out, however, that we observed an interesting phenomenon. In YM64/pMUX6 and YM44/pMUX6, the MICs of several antimicrobial agents were lower than the controls (Table 3). pMUX6 carries muxABC but not opmB (Table 1). This means that overexpression of MuxABC without OpmB conferred elevated susceptibilities to several antimicrobial agents in cells of YM64 and YM44. One possible reason is that overproduced MuxABC may recruit an OMP component from another functional efflux pump (or pumps). It seems unlikely that MuxABC and the recruited OMP would exhibit drug-resistance activity. Meanwhile, the efflux system that has lost its OMP component would no longer function as a pump. It is likely that possible candidates include certain efflux pumps from the MF family. EmrAB, an MF family multidrug efflux pump from E. coli, has been reported to function with the OMP component TolC (Lewis, 2000), and genes for its orthologue are present in the genome of P. aeruginosa (PA5159-PA5160). We are now investigating the function of all the deduced MF-type multidrug efflux pumps in P. aeruginosa. Other candidates are also possible.

In this report, we have characterized the last RND-type multidrug efflux pump predicted from the genome sequence in P. aeruginosa. All the putative RND-type multidrug efflux pumps predicted from the genome sequence have been shown to be responsible for resistance to antimicrobial agents or metal ions. Thus, the primary characterization of all RND-type efflux pumps in P. aeruginosa PAO1 has been completed. We now have some understanding of the substrate specificities for all of the P. aeruginosa RND pumps and of their contributions to intrinsic resistance to antimicrobial agents in P. aeruginosa. It should be noted, however, that such properties of each of the RND pumps became clear only under laboratory conditions. The expression patterns of each efflux pump in biofilm or in the human body might be different from those observed under laboratory conditions. It would thus be important to know the expression profiles and properties of each pump in biofilm or in the human body, where treatment of infectious disease by P. aeruginosa becomes necessary.

We thank Dr Herbert P. Schweizer of Colorado State University for providing us with E. coli strain SM10 lacI^q and plasmids pEX100T, pPS858, pUCP20T and pUCP21T. We thank Dr Manuel Varela of Eastern New Mexico University for critically reading the manuscript. This research was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Edited by: A. R. Walmsley

Footnotes

^,‡, Naoki Kohira^{1 ,†}, Yang Li¹, Hiroshi Sekiya¹ †These authors contributed equally to this work. ‡Present address: Department of Microbiology, Immunology and Pathology, Colorado State University, Fort Collins, CO 80523-1682, USA. §Present address: Department of Infectious Diseases, College of Pharmaceutical Sciences, Matsuyama University, Bunkyo-cho, Matsuyama, Ehime, 790-8578, Japan.