ABSTRACT
One way to combat multidrug-resistant microorganisms is the use of efflux pump inhibitors (EPIs). Spontaneous mutants resistant to the EPI reserpine selected from Streptococcus pneumoniae NCTC 7465 and R6 at a frequency suggestive of a single mutational event were also multidrug resistant. No mutations in pmrA (which encodes the efflux protein PmrA) were detected, and the expression of pmrA was unaltered in all mutants. In the reserpine-resistant multidrug-resistant mutants, the overexpression of both patA and patB, which encode ABC transporters, was associated with accumulation of low concentrations of antibiotics and dyes. The addition of sodium orthovanadate, an inhibitor of ABC efflux pumps, or the insertional inactivation of either gene restored wild-type antibiotic susceptibility and wild-type levels of accumulation. Only when patA was insertionally inactivated were both multidrug resistance and reserpine resistance lost. Strains in which patA was insertionally inactivated grew significantly more slowly than the wild type. These data indicate that the overexpression of both patA and patB confers multidrug resistance in S. pneumoniae but that only patA is involved in reserpine resistance. The selection of reserpine-resistant multidrug-resistant pneumococci has implications for analogous systems in other bacteria or in cancer.
Efflux pumps in bacteria contribute to intrinsic resistance to a wide range of antibiotics and often have a broad substrate range. Typically, the overexpression of multidrug efflux pumps confers resistance to antibiotics, including fluoroquinolones, some dyes (e.g., ethidium bromide), detergents (e.g., sodium dodecyl sulfate [SDS]), and disinfectants (e.g., cetrimide) (41). Genes encoding efflux pumps are typically chromosomally carried, ensuring that mutations are retained in further generations and will not be lost in the absence of selective pressure. Streptococcus pneumoniae is an important cause of community-acquired respiratory infections, including sinusitis, otitis media, and pneumonia, as well as serious invasive infections such as septicemia and meningitis (14). Antibiotic resistance to β-lactams, macrolides, quinolones, and tetracycline is an increasing problem in many countries (6, 27, 46). It has been suggested previously that efflux pumps are a requirement for the selection of fluoroquinolone resistance in S. pneumoniae (25). The overexpression of multidrug efflux pumps can lead to low-level multidrug resistance, which poses a clinical problem (42). Due to the lack of new antibacterial agents, there is considerable interest in restoring the activities of older antibiotics. One way to do this is to inhibit the actions of efflux pumps, and this is an area of active drug development by pharmaceutical companies (29). However, for many species, little is known about the interactions of such inhibitors with the bacterium or even the nature of the inhibitor target protein(s).
The plant alkaloid reserpine was shown previously to be an inhibitor of the Bmr efflux pump of Bacillus subtilis (1), and although rarely used clinically due to toxicity issues, this agent has been widely used in in vitro studies of the activities of new antibacterial agents, particularly fluoroquinolones tested against S. pneumoniae (8, 11, 12, 37). It has been concluded previously that the reduction of the MICs of fluoroquinolones by reserpine indicates an active efflux system (8, 11, 12, 37). In addition, several mutant S. pneumoniae strains with phenotypes suggesting enhanced efflux have been described previously (7, 8, 21); the multidrug resistance displayed was eliminated by reserpine, and this result was taken to indicate the involvement of an active efflux system.
Until recently, the only pneumococcal efflux pump implicated in fluoroquinolone efflux and resistance was PmrA (18). However, multidrug-resistant S. pneumoniae isolates carrying a nonfunctional pmrA gene have been described previously (13, 40). Marrer et al. (31, 32) described two putative ABC efflux pumps, PatA (SP2075) and PatB (SP2073), involved in multidrug resistance in a laboratory-selected ciprofloxacin-resistant mutant. In parallel, Robertson et al. (45) inactivated 13 genes encoding putative efflux pumps in S. pneumoniae R6. The inactivation of the SP2073 and SP2075 genes gave rise to hypersusceptibility to ciprofloxacin and norfloxacin, as well as ethidium bromide and acriflavine. Interestingly, the two mutants were also hypersusceptible to the plant alkaloid berberine.
There are little biochemical data on the proteins with which reserpine interacts, and to date, little evidence that reserpine interacts as an inhibitor of specific efflux pump proteins in S. pneumoniae has been provided. In the present study, it was hypothesized that reserpine interacts directly with a pneumococcal protein and that a reserpine-resistant mutant could be selected. The analysis of such a mutant would further the understanding of the interaction of reserpine with S. pneumoniae and, should the protein be an efflux pump, provide data for rational pneumococcal efflux pump inhibitor (EPI) design. Therefore, reserpine-resistant mutants from two strains of S. pneumoniae were selected and characterized. The mutants were multidrug resistant, and the substrate profile included clinically relevant antibiotics such as fluoroquinolones. Analysis revealed that the increased expression of the genes encoding the two ABC transporters PatA and PatB was associated with multidrug resistance but that only one of these transporters, PatA, was associated with reserpine resistance. This is the first report to show that reserpine can select multidrug-resistant bacteria.
MATERIALS AND METHODS
Bacterial strains, storage, and growth.Eighteen strains of S. pneumoniae were used in this study (Table 1). S. pneumoniae was grown in brain heart infusion (BHI) broth (Oxoid, Basingstoke, United Kingdom) for 24 h at 37°C in 5% CO2. The identification of each species was confirmed by Gram staining and optochin sensitivity testing, and the presence of a capsule was determined by using a Slidex Pneumo-Kit (Biomerieux, France). The growth kinetics of all the strains alone or in the presence of reserpine were measured by assessing the optical densities and by determining the total counts of viable cells over time.
Bacterial strains
Antibiotics and susceptibility determination.The MIC of each antibiotic, dye, detergent, disinfectant, and EPI for each strain was determined by the agar and broth doubling microdilution methods according to the guidelines of the British Society for Antimicrobial Chemotherapy (3). The following antibiotics, dyes, detergents, disinfectants, and EPIs were made up and used according to the instructions of the indicated manufacturers: chloramphenicol, ciprofloxacin, erythromycin, fusidic acid, kanamycin, norfloxacin, novobiocin, penicillin, spectinomycin, tetracycline, vancomycin, acriflavine, ethidium bromide, SDS, cetrimide, carbonyl cyanide m-chlorophenylhydrazone (CCCP), reserpine, sodium orthovanadate, valinomycin, and verapamil (Sigma-Aldrich Company Ltd., Dorset, United Kingdom); ceftriaxone (Roche DPC Europe, Paris, France); levofloxacin (Hoechst Marion Roussel, Strasbourg, France); and moxifloxacin (Bayer AG, Leverkusen, Germany).
Selection of reserpine-resistant mutants. S. pneumoniae strains NCTC 7465 (M4) and R6 (5) were grown at 37°C in 5% CO2 until the optical density at 550 nm reached 0.6. Iso-Sensitest agar plates supplemented with 5% defibrinated horse blood containing no reserpine or two, four, six, or eight times the MIC of reserpine for each strain were inoculated with bacterial suspensions containing between 105 and 109 CFU/ml obtained by centrifuging and concentrating the cells or diluting them in sterile BHI broth (44). The plates were incubated at 37°C in 5% CO2 for 1 to 7 days and examined daily for colonies. At the highest reserpine concentration, any colonies (putative mutants) growing on the reserpine-containing medium with the same colonial morphology as the parent strain were subcultured to antibiotic-free sterile BHI broth. Once the culture had grown to an optical density at 550 nm of 0.6, the mutants were screened for reserpine resistance by agar dilution MIC testing. The frequency of mutation was calculated by dividing the number of mutants isolated at a particular concentration of reserpine by the count of viable cells in the inoculum subcultured in parallel on antibiotic-free Iso-Sensitest agar supplemented with 5% defibrinated horse blood. Mutant selection was also carried out with four different strains of S. pneumoniae to see if reserpine-resistant mutants could also be obtained from other wild-type strains and clinical isolates (Table 1). Reserpine resistance was transformed to S. pneumoniae R6 as described by Marrer et al. (31).
Concentrations of ciprofloxacin and ethidium bromide accumulated by S. pneumoniae.The accumulation of ciprofloxacin by six strains of S. pneumoniae (R6, M169, M184, M4, M168, and M186 [Table 1]) was measured by a fluorometric uptake assay essentially as described previously (39). The concentrations of ethidium bromide accumulated were measured by the fluorescence method (36). The effects of the EPIs CCCP (an uncoupling agent), reserpine (a competitive EPI), verapamil (a competitive inhibitor of MDRI P-glycoprotein and also an ABC transporter and a calcium blocker), valinomycin (an agent that can abolish membrane potential), and sodium orthovanadate (an inhibitor of ATPase) on ciprofloxacin and ethidium bromide accumulation were also determined. All accumulation experiments were performed on at least three separate occasions, and comparisons of the concentrations accumulated in different experiments were analyzed by a two-tailed Student t test. A P value of <0.05 was considered significant.
Expression of efflux pump genes pmrA, patA, and patB, the SP2077 transcriptional regulator gene, and the mismatch repair gene hexA.To measure the levels of expression of pmrA, patA, patB, hexA, and the SP2077 gene from a single mRNA preparation in parallel, the technique of comparative reverse transcription (RT)-PCR (C-RT-PCR) was combined with the rapid and high-throughput technique of denaturing high-pressure liquid chromatography analysis of amplimers as described previously (17, 34, 38, 48). To determine whether there was a significant difference in the mean peak areas indicating the expression of pmrA, patA, patB, hexA, and the SP2077 gene among strains R6, M169, M184, M4, M168, and M186, a two-tailed Student t test with a 5% significance level was used.
PCRs and sequencing of pmrA, patA, patB, hexA, and the SP2077 gene.Two sets of specific primers for pmrA were made to amplify this gene for sequencing, while only one set of specific primers each for patA, patB, hexA, and the SP2077 gene was made to amplify these genes for sequencing (Table 2). Sequencing was performed with the BigDye Terminator version 3.1 cycle sequencing kit (Applied Biosytems Ltd., United Kingdom), and the results were read on the ABI Prism 3700 DNA analyzer.
Details of the DNA primers used in this investigation
In vitro mariner transposon mutagenesis.Plasmid pr412 containing a gene conferring spectinomycin resistance was used as a source of the magellan2 minitransposon essentially as described by Martin et al. (33). Transposition reactions were performed using Himar1 transposase as described previously (23). Targets for transposition were S. pneumoniae strain M169, M184, or M186 DNA (or patA or patB PCR amplimers). Gaps in transposition products were repaired as described previously (2), and S. pneumoniae was transformed with repaired transposition products as described above. The only exceptions were where M169, M184, or M186 was the recipient for transformation instead of R6 and any putative transformants were selected on Columbia blood agar plates containing 100 μg of spectinomycin/ml instead of reserpine. Two test PCRs were performed to confirm the insertion of the magellan2 minitransposon into the target PCR amplimer (patA or patB). For instance, to determine if the magellan2 minitransposon had inserted into patA, two PCRs were performed. The first PCR used the forward primer specific for the amplification of patA and the primer specific for the magellan2 minitransposon inverted terminal repeats (primer set 1) (Table 2), and the second PCR used the reverse primer specific for the amplification of patA and the primer specific for the minitransposon (primer set 2) (Table 2). The resulting amplimers were cleaned, quantified, and sequenced as described above.
Structural analysis of PatA and PatB.With the sequence data generated, Internet tools and Web-based programs were used to analyze the DNA sequences of the efflux pump genes pmrA, patA, and patB of S. pneumoniae and the corresponding amino acid sequences. DNA and amino acid sequences were analyzed using GCG (Genetics Computer Group, University of Birmingham) and imported into GeneDoc (http://www.nrbsc.org/downloads/ ) for sequence alignment. By using Web-based programs such as SOSUI (http://bp.nuap.nagoya-u.ac.jp/sosui/sosui_submit.html ), the peptide sequence of either PatA or PatB was analyzed. SOSUI predicts which peptides reside in the transmembrane regions of efflux pumps. The peptide sequences generated from the sequencing reactions for the wild-type strains of S. pneumoniae were entered into the program, and the results from the program were used to generate two-dimensional structures of the efflux pumps demonstrating the conformation in which each peptide lies. The hydrophobicity profiles of either PatA or PatB in the wild-type strains of S. pneumoniae were determined by using TMHMM version 2.0 (http://www.cbs.dtu.dk/services/TMHMM-2.0/ ). This program predicts the transmembrane arrangements of proteins. SWISS-MODEL and Protein Explorer (http://swissmodel.expasy.org/SWISS-MODEL.html ) were used to generate putative structures of either PatA or PatB based on similar homologues for which a structure had been resolved previously. BLAST and Pfam searches for close homologues of PatA and PatB were conducted.
RESULTS
Reserpine-resistant mutants of S. pneumoniae are easily selected. S. pneumoniae strains R6 and M4 (capsulated type 2; NCTC 7465) were exposed to 128 μg of reserpine/ml (double the MIC for each strain) on at least two separate occasions. When mutants were obtained, the frequencies of mutation to resistance were 7.3 × 10−6 and 3.59 × 10−5, respectively, suggesting a mutation in a single gene. Two reserpine-resistant mutants of R6 and M4, M169 and M168, respectively, were randomly chosen and retained for further investigation. Due to the high frequencies of mutation to resistance shown by R6 and M4, mutant selection with reserpine was performed on two separate occasions with four more strains of S. pneumoniae, including another type strain and three clinical isolates. When mutants were selected, the frequency of mutation to resistance was also high, between 4.0 × 10−5 and 7.3 × 10−6. Both reserpine-resistant mutants M169 and M168 grew as well as their respective parent strains R6 and M4 (generation time for M169, 25.7 ± 4.8 min, versus 25 ± 1.6 min for R6, and generation time for M168, 28.2 ± 3.2 min, versus 28.3 ± 4.8 min for M4).
Reserpine resistance is mediated by a single mutational event.Reserpine resistance was transferred by natural transformation from M168 and M169 to R6. This transfer allowed the mechanism of reserpine resistance to be characterized in a clean background. R6 is also easier to manipulate genetically than the capsulated strain M4. With M168 chromosomal DNA, a transformation frequency of 1.8 × 10−5 was obtained; 180 transformants examined showed a susceptibility pattern identical to that of M168 (Table 3). Transformant M186 was randomly chosen for further study. With M169 chromosomal DNA, a transformation frequency of 2.48 × 10−5 was obtained; the 248 transformants examined showed a susceptibility pattern identical to that of M169 (Table 3). Transformant M184 was randomly chosen for further study. No spontaneous reserpine-resistant mutants were selected using a negative control containing water instead of DNA.
Activities of antibiotics, dyes, detergents, and reserpine for the strains, mutants, and transformants in this studya
Reserpine selects for multidrug resistance.The wild-type S. pneumoniae strains R6 and M4 showed the typical susceptibilities of S. pneumoniae to all agents tested (Table 3). The reserpine-resistant mutants M168 and M169 were fourfold more resistant to reserpine than the isogenic parent strains M4 and R6. M168 and M169 were also multidrug resistant. Both mutants were four- to sixfold less susceptible to the dyes ethidium bromide and acriflavine and four- to eightfold less susceptible to the fluoroquinolone antibiotics ciprofloxacin, norfloxacin, and levofloxacin and the disinfectant cetrimide (Table 3). The minimum bactericidal concentrations of fluoroquinolones, cetrimide, ethidium bromide, acriflavine, and reserpine for M168 and M169 were fourfold higher than those for M4 and R6, respectively (data not shown).
pmrA is not involved in reserpine resistance or multidrug resistance.To determine whether the region of pmrA homologous to the region encoding the putative reserpine-binding pocket of Bmr (22) contained a mutation that conferred an amino acid substitution and hence reserpine resistance, the DNA sequences were determined. The nucleotide sequences of the pmrA genes of M169 and M184 were identical to that of the pmrA gene of R6. The sequencing of pmrA genes of M4 and M168 confirmed that both strains contained a 28-amino-acid deletion situated between amino acids 158 and 186 and thereby including transmembrane-spanning region 6, as seen previously (32). However, DNA sequencing revealed one mutation in M168 pmrA compared to M4 pmrA, at nucleotide (nt) 127 (cytosine to guanidine), leading to the replacement of the amino acid glutamine with similarly sized but negatively charged glutamic acid. However, the nucleotide sequence of pmrA of the transformant M186 was the same as that of pmrA of R6.
The level of expression of pmrA in M4 was 2.3- ± 0.7-fold lower than that of pmrA in R6 (P = 0.007). The levels of expression of pmrA in M4 and M168 were not significantly different (P = 0.63); this was also the case for pmrA in R6 and M169 (P = 0.93) (data not shown). The levels of expression of pmrA in R6 and the transformant strains M184 (P = 0.40) and M186 (P = 0.30) were similar (data not shown).
Sodium orthovanadate abolishes both the efflux of and resistance to ciprofloxacin and ethidium bromide.The reserpine-resistant mutants M169 and M168 were also multidrug resistant, suggesting an efflux mutant phenotype. As decreased susceptibilities of both mutants to ciprofloxacin and ethidium bromide were observed, the accumulation and efflux of both agents were measured. The reserpine-resistant mutant M169 and its transformant M184 accumulated significantly less ciprofloxacin (2.9- and 2.8-fold less, respectively) than R6 (P = 0.0001 and 0.008, respectively) (Fig. 1). M4, the capsulated parent strain, accumulated less ciprofloxacin (1.6-fold) than R6 (Fig. 1). M168 accumulated significantly less ciprofloxacin (2.3-fold) than M4 (P = 0.0002); M186 also accumulated less (2.1-fold) than R6 (P = 0.007) (Fig. 1).
Accumulation of ciprofloxacin (CIP) with or without reserpine or sodium orthovanadate in S. pneumoniae strains R6, M169, M184, M186, M4, and M168. M186 was compared to R6 because M186 is a mutant of R6 transformed with DNA from M168. An asterisk indicates a P value of <0.05. R, reserpine; O, sodium orthovanadate.
The wild-type strain R6 accumulated more ethidium bromide over 10 min than the reserpine-resistant mutant M169 and its transformant M184 (Fig. 2A). After 10 min, M169 and M184 accumulated 24.6 and 26.7% less ethidium bromide, respectively, than R6 (P = 0.016 and 0.035, respectively) (Fig. 2A). Likewise, M4 accumulated more ethidium bromide than its reserpine-resistant mutant M168, with M168 accumulating 27.7% less ethidium bromide than M4 (P = 0.029) (Fig. 2B). M186 accumulated 32.9% less ethidium bromide after 10 min than R6 (P = 0.034) (Fig. 2B). To give an indication as to which type of pump was responsible for the reduced accumulation observed, accumulation in the presence of various inhibitors of efflux, including CCCP, valinomycin, verapamil, reserpine, and sodium orthovanadate, was measured. Only sodium orthovanadate inhibited the efflux of ciprofloxacin and ethidium bromide and increased the concentrations of these agents accumulated (Fig. 1 and 2) (data for the other inhibitors are not shown). Sodium orthovanadate inhibits the action of an efflux pump (or any other system) that uses ATP hydrolysis as an energy source (19, 24, 49).
Accumulation of ethidium bromide with or without sodium orthovanadate in S. pneumoniae. (A) Ethidium bromide kinetics over 10 min for strains R6 (⧫), M169 (□), M169 in the presence of sodium orthovanadate (▪), M184 (○), and M184 in the presence of sodium orthovanadate (•). Sodium orthovanadate was added after 5 min (as indicated by the arrow). (B) Ethidium bromide kinetics over 10 min for strains M4 (⧫), M168 (□), M168 in the presence of sodium orthovanadate (▪), M186 (○), and M186 in the presence of sodium orthovanadate (•). Sodium orthovanadate was added after 5 min (as indicated by the arrow).
The effect of inhibitors of efflux upon antimicrobial activity was also determined (Table 3). Reserpine (20 μg/ml) and sodium orthovanadate (25 μM) reduced the MIC of ciprofloxacin by twofold for the parent strains and reserpine-resistant mutants (Table 3). Sodium orthovanadate lowered the MICs of acriflavine more than those of reserpine, with the greatest reductions in MICs for the mutants and transformants (8- to 10-fold). The effects of higher concentrations of reserpine revealed that 80 μg of reserpine/ml reduced the MIC of ciprofloxacin for M169, M184, and M186 to 1 μg/ml, the same as that for R6. However, 40 μg of reserpine/ml reduced the MICs of ethidium bromide and acriflavine to the same levels as those for the wild type (R6).
No radiolabeled reserpine was available to determine whether reserpine was a substrate of PatA and/or PatB.
patA and patB are overexpressed in the reserpine-resistant mutants.As the accumulation data suggested that an ABC transporter was involved in ciprofloxacin and ethidium bromide resistance, the expression of the genes encoding two ABC transporters previously implicated in ciprofloxacin resistance (32), patA and patB, was measured by C-RT-PCR. The expression of patA and patB (and pmrA) was constitutive and consistently detected at 1,000-fold-lower levels than that of 16S rRNA. The levels of expression of patA in the mutants and transformants were significantly higher than those in the parent strains; M169 and M184 expressed patA at 1.3- ± 0.2-fold- and 1.3- ± 0.3-fold-higher levels than R6 (Fig. 3). The level of expression of patB was also higher in M169 (1.4- ± 0.1-fold) and M184 (1.5- ± 0.1-fold) than in R6 (Fig. 3). The levels of expression of patA and patB were higher in M168 and M186 (1.4- ± 0.2-fold and 1.3- ± 0.2-fold for patA, respectively, and 1.3- ± 0.1-fold and 1.5- ± 0.2-fold for patB, respectively) than in M4 (Fig. 3).
Comparison of the levels of expression of patA (□) and patB (▪). The asterisks indicate values significantly different (P < 0.05) from those for R6.
patA and patB encode ABC transporters.The amino acid sequences of PatA and PatB were analyzed in order to predict the structure of each protein. Pfam and BLAST homology searches indicated that both patA and patB genes encode ABC transporters. Both PatA and PatB contain the signature motif, Q loop, and H loop/switch region in addition to the Walker A motif/P loop and Walker B motif commonly found in ATP- and GTP-binding and -hydrolyzing proteins (20, 43, 50). SWISS-MODEL was used to generate putative structures of each protein; however, only PatB had strong enough homology to a known protein (Sav 1866 of Staphylococcus aureus) to give a structure. Again, this structure was indicative of a typical ABC homodimeric transporter and showed six transmembrane domains (see Fig. S1 in the supplemental material). Sav 1866 had 34% amino acid identity and 54% amino acid similarity to PatB. PatA had 29% amino acid identity and 53% amino acid similarity to Sav 1866, and these results fell just below the threshold for SWISS-MODEL to generate a structure of PatA based on Sav 1866. PatA had 27% amino acid identity and 52% amino acid similarity to PatB. SOSUI, a program which generates a two-dimensional structure of the protein and demonstrates the conformation in which each peptide lies, predicted that PatA contains six transmembrane domains with two large external loops, indicative of a typical ABC transporter. The hydrophobicity profile of PatA was determined by using TMHMM version 2.0. This program predicts the transmembrane arrangements of proteins and predicted PatA to contain six transmembrane domains.
BLAST searches revealed that PatA and PatB have close homologues that always occur as linked pairs in gram-positive bacteria (Table 4 and see Fig. S2 in the supplemental material). An Entrez PubMed protein BLAST search (http://www.ncbi.nlm.nih.gov/BLAST/BLAST.cg ) seeking proteins with as low as 35% amino acid identity to PatA or PatB revealed similar pairs of ABC transporters in 42 bacterial species. Among the homologues identified were two systems that have been implicated previously in efflux-mediated multidrug resistance in Enterococcus faecalis (EfrA and EfrB) (24) and Lactococcus lactis (LmrC and LmrD) (30) and three systems that are part of biosynthetic pathway operons for the polyene macrolactone antibiotics nystatin (NysG and NysH) (10), pimaricin (PimA and PimB) (4), and amphotericin (AmphG and AmphH) (16). All of these bacterial ABC transporters are thought to function as heterodimers or as pairs in which both proteins work together as a multidrug efflux pump (9). The coding sequences for the homologues in Streptomycetes overlap, suggesting that the two components of the transport system may be translationally coupled (4, 10, 16). The homologues of PatA and PatB in Lactococcus (LmrC and LmrD) are interdependent for their activity in drug resistance (30), consistent with the model of a pump comprising two proteins. patA and patB are in close proximity to one another on the pneumococcal genome and are separated by SP2074, a pseudogene thought to encode a transposase (Fig. 4). Most of the literature on close homologues of PatA and PatB (Table 4) and bioinformatic data suggest that homologues of PatA and PatB function together. In light of this finding, it is predicted that PatA and PatB form a heterodimeric efflux pump.
Locations of SP2070, SP2071, SP2072, SP2073 (patA), SP2074, SP2075 (patB), SP2076, and SP2077 genes. patA and patB are colocalized in the S. pneumoniae (TIGR4) genome.
Homologues of PatA and PatB found in bacteria other than S. pneumoniae
The inactivation of patA confers a loss of multidrug resistance and reserpine resistance, while the inactivaton of patB confers a loss of multidrug resistance only.To determine whether the increased expression of patA and/or patB conferred reserpine resistance and multidrug resistance, in vitro mariner mutagenesis was used to inactivate patA and patB in the mutants and transformants. DNA sequencing revealed that the magellan2 minitransposon had inserted between an AT nucleotide pair at nt 833 and 834 of patA in all strains and had inserted between an AT nucleotide pair at nt 1082 and 1083 of patB. Irrespective of the strain, the inactivation of patA or patB conferred hypersusceptibility to the fluoroquinolones, acriflavine, and ethidium bromide (Table 3). For both M246 (R6 patA::magellan2) and M240 (R6 patB::magellan2), there was no change in reserpine susceptibility compared to that of R6 (Table 3). DNA sequencing revealed that patA and patB were inactivated, as in R6. However, M230, M233, and M235 (all patB::magellan2 strains) remained fourfold less susceptible to reserpine than R6, whereas for M231, M234, and M236 (all patA::magellan2 strains), the MIC of reserpine was reduced to the level seen for the wild-type R6 (Table 3). Reserpine had no effect on the MICs of any of the agents tested for any strains with inactivated patA and/or patB.
M246 (R6 patA::magellan2) grew more slowly than R6 (generation time for M246, 56.67 ± 9.95 min; generation time for R6, 36.24 ± 2.85 min; P = 0.0046). However, the growth rate of M240 (R6 patB::magellan2; generation time, 34.37 ± 2 min) was the same as that of R6 (see Fig. S3A in the supplemental material). The growth kinetics of the patA and patB insertionally inactivated strains of the reserpine-resistant mutants and transformants also indicated that patA insertionally inactivated strains grew more slowly than the wild type (see Fig. S3B in the supplemental material).
In M246 (R6 patA::magellan2), patB was overexpressed 1.85-fold compared to patB expression in R6 (P = 0.003), but in M240 (R6 patB::magellan2), the expression of patA was unaltered compared to that in R6.
No mutations were found in patA or patB or upstream regions.To determine whether the decreased accumulation of ciprofloxacin and ethidium bromide was due to a functional change in the transporter protein due to an amino acid substitution, the nucleotide sequences of the patA and patB genes were determined. Those of M169, M184, and M186 were identical to those of R6, and those of M168 were identical to those of M4.
In light of the results implying that the overexpression of patA (and patB) was responsible for reserpine resistance and multidrug resistance, the upstream region encompassing the putative promoter region (∼200 bp from the transcriptional start codon) of patA and patB was sequenced. No mutations in M169, M184, M168, or M186 were detected.
patA and patB are overexpressed in clinical isolates.To determine whether the overexpression of patA and/or patB was associated with multidrug resistance in clinical isolates of S. pneumoniae and so is a clinically relevant mechanism of antibiotic resistance, the expression of patA and patB in 18 multidrug-resistant clinical isolates was assessed. These isolates had been previously characterized; none overexpressed pmrA, but reserpine abolished the multidrug resistance (40). Five of the isolates significantly overexpressed both patA (1.2- to 1.7-fold) and patB (1.2- to 1.6-fold) compared to wild-type susceptible strains. In addition, S. pneumoniae strain AMC-058 and two isogenic efflux mutants (mutants with multidrug resistance abolished by 10 μg of reserpine/ml) selected in vivo after ciprofloxacin exposure, strains RC2 and RC4 (21), were also studied. C-RT-PCR revealed that both patA and patB were significantly overexpressed in these strains (1.19- to 1.34-fold and 1.13- to 1.38-fold, respectively) compared to the expression in strain AMC-058.
hexA and the SP2077 gene are not involved in reserpine or multidrug resistance.Two genes upstream of patA and patB were identified in the bioinformatic analysis: hexA (a mutS equivalent involved in mismatch repair) and the SP2077 gene (encoding a putative transcriptional repressor) (Fig. 4). To determine whether these genes were involved in reserpine and/or multidrug resistance, the sequences were determined and expression was measured by C-RT-PCR. The nucleotide sequences of the hexA and SP2077 genes of M169, M184, and M186 were identical to those of R6, and those of M168 were identical to those of M4. The levels of expression of the hexA and SP2077 genes in all strains were similar (data not shown). These data also confirm the lack of polar effects of the inactivation of patA and patB.
DISCUSSION
Many studies have suggested that efflux pumps in microorganisms and cancer cells provide an attractive target for inhibition (29). An EPI should increase the activity of an antibiotic for wild-type and multidrug-resistant cells, as efflux pumps provide both innate and higher-level resistance. To date, only one inhibitor (MC-207,110, which is a competitive inhibitor of resistance-nodulation-cell division [RND] pumps of gram-negative bacteria) has been extensively characterized (28), and this characterization indicated significant potential for developing small-molecule inhibitors against efflux pumps. Due to the continued clinical pressure for novel approaches to combat antibiotic-resistant bacterial infections, programs to identify EPIs continue (29). As progress in the development of EPIs proceeds, the present study provides a cautionary note of the possibility that the use of an EPI can select multidrug-resistant bacteria. This is the first report of an EPI selecting a multidrug-resistant strain of bacteria, and therefore, any therapeutic strategy based on an antibiotic and an EPI needs to take into account that the use of an EPI may lead to more resistant strains.
There is little biochemical evidence to indicate the protein(s) with which reserpine interacts. Most literature describes reserpine interacting with the rat chromaffin granule vesicular amine transporter, the mammalian multidrug efflux pump P-glycoprotein, and Bmr of B. subtilis (1, 35). In addition, it needs to be noted that reserpine may also (i) be an alternative substrate for an efflux pump, albeit a poor one for many efflux systems, and (ii) have additional toxic effects on the bacteria, thereby inhibiting growth. In the present study, to identify the target protein of reserpine in pneumococci, spontaneous mutants with decreased susceptibilities to reserpine were selected. The mutants were stably resistant to reserpine and were multidrug resistant. Given the results from studies by Ahmed et al. (1), Klyachko et al. (22), and Gill et al. (18), the role of the efflux pump protein PmrA in decreased susceptibility to reserpine and multidrug resistance was exhaustively explored. Our data indicate that the reserpine resistance and the multidrug resistance seen in the selected mutants were not due to PmrA, as no substitutions within PmrA or overexpression of pmrA were seen in the mutants of S. pneumoniae R6. Although a substitution within PmrA was seen in the mutant of the capsulated strain of S. pneumoniae M4, this strain also contained a 28-amino-acid deletion and showed minimal expression of pmrA; the mutation was not transformed to R6 with the reserpine resistance and multidrug resistance.
To elucidate whether an efflux system was responsible for the multidrug resistance and reserpine resistance, accumulation experiments were performed with ciprofloxacin and ethidium bromide (in the presence and absence of different inhibitors of efflux). The accumulation data indicated that an efflux pump system was involved in the resistance to ciprofloxacin and ethidium bromide and that the pump was an ABC transporter. Although data obtained with sodium orthovanadate suggest that an ABC transporter is involved, it must be noted that in any experiment performed with intact cells and not purified protein(s), there is a possibility that the effect seen results from interaction with a protein specific to sodium orthovanadate and not interaction with an efflux protein. Our previous work implicated two putative ABC transporters, PatA and PatB, in ciprofloxacin resistance and multidrug resistance in pneumococci (31, 32). The inactivation of patA and patB in R6 also confers hypersusceptibility to several agents (45). Irrespective of whether the strain was a reserpine-resistant multidrug-resistant laboratory mutant or a multidrug-resistant clinical isolate, products of both genes were always overproduced compared to those in susceptible strains, as indicated by the inactivated mutants. Despite the relative modesty of this overexpression, it was sufficient to confer a fourfold decrease in susceptibility to reserpine. This modest increase in the expression of patA and patB resulting in reserpine resistance and multidrug resistance is reminiscent of the relatively small increase in the expression of mexAB-oprM in Pseudomonas aeruginosa, which also translates into measurable resistance (with the MIC for the bacterium increasing two- to sixfold compared to that for strains without overexpression) (26, 47). In addition, patA and patB were always overexpressed together, suggesting that the ciprofloxacin and ethidium bromide resistance is due to enhanced efflux via these putative ABC transporters. Bioinformatic data further supported the hypothesis that PatA and PatB are ABC transporters and also revealed that homologues always occur as linked pairs in gram-positive bacteria. The majority of these pairs function as heterodimeric multidrug efflux pumps (9).
As Klyachko et al. (22) suggested that mutations within an efflux pump gene could confer altered susceptibility to reserpine, the sequences of patA and patB in all the strains were determined. However, no amino acid substitutions in PatA and PatB were revealed. Both the reserpine-resistant mutants and transformants overexpressed patA and patB, suggesting that enhanced efflux confers multidrug resistance. Through in vitro mariner mutagenesis inactivation of patA and patB, we further explored the roles of PatA and PatB in reserpine resistance. When patA and patB in the reserpine-resistant mutants were disrupted, the multidrug resistance was abolished and the levels of accumulation of ciprofloxacin and ethidium bromide were restored to wild-type levels, suggesting that the overexpression of both patA and patB was responsible for multidrug resistance. However, only in the patA-disrupted mutants was the reserpine resistance lost, suggesting that reserpine interacts with PatA only. It may be that PatA forms a homodimer and is responsible for reserpine resistance and that a heterodimer consisting of PatA and PatB is responsible for multidrug resistance. In addition, when patA was insertionally inactivated, growth was slower than that of the wild type, confirming the observation of Marrer et al. (32). Interestingly, when patA in R6 was disrupted, patB was overexpressed, but not vice versa.
This study provides data to support a role for PatA and PatB as ABC transporters and indicates that they are involved in multidrug resistance in S. pneumoniae. Data obtained in this study also suggest that reserpine interacts with PatA and not PmrA as originally thought. ABC transporters are a conserved group of proteins that confer not only multidrug resistance in bacteria but also resistance to anticancer agents in mammalian cells. The development of EPIs is an active area of drug development to improve and/or extend the clinical utility of antibiotics and anticancer agents. If an EPI is a substrate of the same efflux pump as the agent with which it has been combined, common use may drive the development of EPI resistance and also cross-resistance to antibiotics. Data from this study will inform those involved in the development of EPIs and aid inhibitor and drug design.
ACKNOWLEDGMENTS
This work was supported by a Bristol-Myers Squibb unrestricted grant in infectious diseases to L.J.V.P.
We thank Mark Prudhomme for his helpful advice with in vitro transposon mariner mutagenesis experiments and the Society for General Microbiology for funding the visit of Mark I. Garvey to the Prudhomme laboratory for 1 month. We also thank Adrian Walmsley, Mark Webber, and Andrew Bailey for reading the manuscript and for their helpful comments, Sumera Karim for performing growth kinetics analyses, and Kin Long Ryan Wong for evaluating the expression of patA and patB in multidrug-resistant clinical isolates of S. pneumoniae. Thanks also to Roy Chaudhuri for helping us to generate the phylogenetic trees.
FOOTNOTES
- Received 21 December 2007.
- Returned for modification 25 January 2008.
- Accepted 29 February 2008.
- Copyright © 2008 American Society for Microbiology