|Home | About | Journals | Submit | Contact Us | Français|
The outer membrane of mycobacteria presents an effective permeability barrier for many antibiotics. Transport pathways across this membrane are unknown for most drugs. Here, we examined which antibiotics utilize the porin pathway across the outer membrane of the model organism Mycobacterium smegmatis. Deletion of the porins MspA and MspC drastically increased the resistance of M. smegmatis ML10 to β-lactam antibiotics, while its β-lactamase activity remained unchanged. These results are consistent with the ninefold-reduced outer membrane permeability of the M. smegmatis porin mutants for cephaloridine and strongly indicate that β-lactam antibiotics rely on the porin pathway. The porin mutant ML10 accumulated less chloramphenicol and norfloxacin and was less susceptible to these antibiotics than wild-type M. smegmatis. These results demonstrated that small and hydrophilic antibiotics use the Msp porins for entering the cell. In contrast to norfloxacin, the hydrophobic moxifloxacin was 32-fold more effective in inhibiting the growth of M. smegmatis, presumably because it was able to diffuse through the lipid membrane. Structural models indicated that erythromycin, kanamycin, and vancomycin are too large to move through the MspA channel. This study presents the first experimental evidence that hydrophilic fluoroquinolones and chloramphenicol diffuse through porins in mycobacteria. Thus, mutations resulting in less efficient porins or lower porin expression levels are likely to represent a mechanism for the opportunistic pathogens M. avium, M. chelonae, and M. fortuitum, which have Msp-like porins, to acquire resistance to fluoroquinolones.
Mycobacterium tuberculosis has infected about 2 billion people and causes the death of about 2 million people every year, more than any other bacterial pathogen (20). Chemotherapy of tuberculosis (TB) is effective in patients infected with susceptible M. tuberculosis strains. However, infections with multidrug-resistant and extensively drug-resistant strains of M. tuberculosis are increasing throughout the world (51). M. tuberculosis quickly becomes resistant in therapeutic regimens based on a single drug (31, 57) due to target mutations (56). In addition to these acquired resistance mechanisms, mycobacteria, including M. tuberculosis, are intrinsically resistant to many antibiotics and drugs. This has been attributed to the low permeability of the cell envelope (3), which acts in combination with drug-inactivating, target-modifying enzymes and multidrug efflux pumps (10, 16, 24, 34). Considering the importance of the few drugs and antibiotics available for TB chemotherapy, it is surprising that the mechanism by which these molecules enter the cell is still unknown.
For a long time, the architecture of the mycobacterial cell wall has been controversial (11). Recently, we demonstrated by cryoelectron microscopy that mycobacteria contain a second lipid bilayer in addition to the inner membrane. Long-chain mycolic acids are essential components of this unusual outer membrane (21). Thus, antibiotics first need to overcome the permeability barrier established by the outer membrane. This requirement explains previous findings that mutants or treatments affecting mycolic acid biosynthesis and the production of extractable lipids show increased permeability and a concomitant increase in susceptibility to hydrophobic drugs (12, 43, 58). In addition, M. tuberculosis mutants with defects in the synthesis of the outer membrane were impaired in virulence, thus underlining the importance of its protective function for intracellular survival (1). Hydrophobic compounds can directly diffuse through lipid membranes, while the permeability of lipid membranes for hydrophilic drugs is extremely low. Therefore, they often utilize channel proteins, such as porins, to cross the outer membranes of bacteria. Antibiotics, such as β-lactams, tetracyclines, fluoroquinolones, chloramphenicol, and cycloserine, have been shown to use porins for access to the periplasm in gram-negative bacteria (37).
To date, only two classes of channel-forming outer membrane proteins are known in mycobacteria: OmpA-like proteins and MspA-like porins in slow- and fast-growing mycobacteria, respectively. OmpA is a channel protein in the outer membrane of M. tuberculosis (50). The ompA mutant of M. tuberculosis does not show increased drug resistance (44), and no other evidence for involvement of OmpA in the uptake of antibiotics is available. MspA is the main porin, constituting more than 70% of all pores of M. smegmatis (52). X-ray analysis of MspA revealed a unique structure with an octameric channel that is more than twice as long as the channel of the porin OmpF and of other porins of Escherichia coli (13). The genes mspB, -C, and -D encode porins that are very similar to MspA (52). Only mspA and mspC are expressed in wild-type (wt) M. smegmatis, while transcription of mspB and mspD is activated upon deletion of mspA by an unknown mechanism (53). Although the uptake rate for glucose was sixfold reduced in the mspA mutant MN01 (52), the growth of M. smegmatis was not much affected (52). Lack of MspA also increased the MICs of ampicillin and cephaloridine for M. smegmatis, indicating that MspA plays an important role in the uptake of hydrophilic β-lactam antibiotics (54). In addition, this study suggested that the MICs of vancomycin, erythromycin, and rifampin were increased for the mspA mutant. This was surprising, because these antibiotics were thought to be too large to diffuse through the MspA pore (54). We hypothesized that M. smegmatis strains with significantly fewer porins than the ΔmspA mutant MN01 should exhibit a stronger resistance phenotype if the increased resistance was indeed due to impaired uptake. To this end, we utilized the porin mutants ML10 (ΔmspA ΔmspC) and ML16 (ΔmspA ΔmspC ΔmspD). The numbers of Msp porins in these mutants are at least 15- and 5-fold lower than in wt M. smegmatis and the ΔmspA mutant MN01, respectively (53). This coincides with 75-fold lower permeability of ML10 and ML16 to nutrient molecules, such as glucose. The growth of both strains was much slower than that of the wt, demonstrating that porin-mediated influx of nutrients is a major determinant of the growth rate of M. smegmatis (53).
In this study we show that the loss of Msp porins does not significantly alter the susceptibility of M. smegmatis to vancomycin, kanamycin, and erythromycin. Further, uptake experiments provide direct evidence that chloramphenicol and fluoroquinolones use Msp porins to cross the outer membrane of M. smegmatis. These findings in the model organism M. smegmatis have important implications for an understanding of how antibiotics can cross mycobacterial outer membranes.
M. smegmatis SMR5 is a derivative of M. smegmatis mc2155 and is streptomycin resistant due to a mutation in ribosomal protein S12 (48). The construction of M. smegmatis porin mutants MN01 (ΔmspA), ML10 (ΔmspA ΔmspC), and ML16 (ΔmspA ΔmspC ΔmspD) was described previously (52, 53). All mycobacterial strains were grown in Middlebrook 7H9 liquid medium (Difco) supplemented with 0.2% glycerol and 0.05% Tween 80 or on Middlebrook 7H10 plates (Difco) containing 0.5% glycerol at 37°C.
Initial drug solutions were prepared in ethanol or distilled deionized water, or in 0.1 M NaOH for fluoroquinolones, and stored at −80°C. MICs for M. smegmatis wt and porin mutants were determined in triplicate by agar dilution experiments on Middlebrook 7H10 agar as described previously (53). A preculture in Middlebrook 7H9 liquid medium (Difco) was filtered through 5-μm filters (Sartorius AG) to remove large cell aggregates and then grown in the same medium to an optical density at 600 nm (OD600) of 0.6 to 0.8. After dilution in 7H9 medium, approximately 500 CFU was streaked on agar plates. For all drugs, the concentration that resulted in the complete inhibition of bacterial growth after 5 days of incubation at 37°C was defined as the MIC. Colonies of the porin mutants ML10 and ML16 were clearly visible after 5 days on Middlebrook 7H10 agar plates.
M. smegmatis wt and porin mutant strains were grown in 4-ml cultures at 37°C until an OD600 of 0.5 to 0.9 was reached. The cultures were diluted to an OD600 of 0.01 to 0.02 in medium, and 100 μl of these dilutions was added per well of black, clear-bottom 96-well microplates (Greiner Bio-One) (27). The microplate Alamar blue assay was performed and analyzed as described previously (7). For all experiments, the fluorometric endpoint was the same within a twofold drug dilution as the visible endpoint, which was determined by a color change from blue to pink upon reduction of Alamar blue by viable bacteria. The slower growth of the porin mutants ML10 and ML16 did not influence the results because the fluorescence intensities were normalized to those of the drug-free control. Aggregation of cells during liquid culture is a known problem for mycobacteria, which is likely to reduce the accessibility and hence the susceptibility of the bacteria to antibiotics. Therefore, all bacterial cultures were examined by phase-contrast microscopy before susceptibility experiments were performed. The vast majority of all cells in these cultures were single cells, and they did not have aggregates larger than approximately 10 cells. Cultures with more or larger clumps were discarded.
Uptake of fluoroquinolones was determined by measuring their intrinsic fluorescence as described previously (8). In a first step, the fluorescences of four fluoroquinolones were examined. Norfloxacin, moxifloxacin, sparfloxacin, and ofloxacin were dissolved in 0.02 M NaOH and diluted in 0.1 M glycine hydrochloride (pH 3.0) to give a final concentration of 0.32 μg/ml. Absorbance and fluorescence spectra were recorded at 25°C using a photometer (Lambda; Perkin Elmer) and a fluorimeter (FP-6500; Jasco). All fluoroquinolones had an absorbance maximum at 283 nm, but only norfloxacin and moxifloxacin showed detectable fluorescence with an emission maximum at 440 nm when excited at 283 nm. The fluorescence of norfloxacin was approximately 100-fold higher than that of moxifloxacin under those conditions.
For the uptake experiments, M. smegmatis cells were harvested by centrifugation at 4°C, washed, and resuspended in sodium phosphate buffer (50 mM; pH 7.0; 0.05% [vol/vol] Tween 80) to 4 × 109 cells per ml. Samples were placed in flasks and preincubated at 37°C for 10 min in a shaking bath for uptake measurements or placed on ice for adsorption assays. Norfloxacin or moxifloxacin was added to a final concentration of 100 μg/ml. Triplicate samples of 0.8 ml were removed at different times and immediately put on ice. Following centrifugation at 16,000 × g (6°C; 1 min), the cells were washed twice with 0.1 M sodium phosphate buffer containing 0.1% (vol/vol) Tween 80 (pH 9.0). The cell pellets were lysed with 1.5 ml of 0.1 M glycine hydrochloride (pH 3.0) overnight at room temperature. The samples were centrifuged at 16,000 × g (20°C; 40 min). The fluorescence of the supernatant at 440 nm was measured after excitation at 283 nm. The fluorescence of control cell lysates that were not treated with fluoroquinolones was subtracted from the fluorescence measured for cells incubated with fluoroquinolones. The cells were dried in vacuum at 60°C for 30 min to determine the dry weight. The results were expressed as μg of fluoroquinolone per mg of cells (dry weight).
The cells were harvested at an OD600 of 0.6 to 1.0 by centrifugation (3,000 × g at 4°C for 10 min), washed twice in uptake buffer [50 mM Tris, pH 6.9, 15 mM KCl, 10 mM (NH4)2SO4, 1 mM MgSO4, 0.1% Tween 80] and resuspended in the same buffer. The cultures were preincubated at 20°C for 5 min before the addition of [14C]chloramphenicol (specific activity, 59 mCi/mmol). The final concentration of the antibiotic in the sample was 20 μM. At the indicated time intervals, 0.2-ml samples were removed and filtered immediately through a 0.45-μm-pore-size filter using a Spin-X centrifuge tube filter (Costar). The cells were washed twice with a killing buffer (1:2 [vol/vol] 0.1 M LiCl/10% formalin). The radioactivity associated with the cells was counted in a liquid scintillation counter (Beckman). The cells were dried in vacuum at 60°C for 30 min to determine the dry weight. The uptake rate was expressed as pmol of antibiotic per milligram of cells.
To determine the total β-lactamase activity in M. smegmatis, the hydrolysis of the cephalosporin antibiotic nitrocefin was measured as previously described (14) with some modifications. M. smegmatis wt and porin mutant strains were grown at 37°C until late logarithmic phase (OD600 ≥ 2.0). To induce β-lactamase expression, 100 μg/ml ampicillin was added to the stationary-phase cultures. The cells were further incubated at 37°C for 2 h to allow protein expression. To obtain cell-free culture filtrates, 500 μl of the culture supernatants was filtered twice through 0.2-μm filters (Pall Corporation). To obtain lysates, cells were pelleted and washed twice in ice-cold 1× phosphate-buffered saline (PBS) buffer (pH 7.0). The cell pellets were resuspended in 1/100 volume of 1× PBS containing corresponding amounts of protease inhibitor cocktail (Sigma) and DNase I (NEB). The cells were disrupted by agitation with glass beads (FastRNA Tubes-Blue) in a FastPrep FP120 bead beater apparatus (Bio-101) twice for 30 seconds each time at level 6.0 with a 5-min rest on ice between agitation steps. Cell debris was removed by centrifugation. The lysates obtained were filtered twice through 0.2-μm filters. Protein concentrations were determined using a bicinchoninic acid protein assay kit (Pierce). Nitrocefin was added to a final concentration of 100 μM in 1× PBS (pH 7.0), and hydrolysis was monitored as a change in absorbance at 490 nm using a microplate reader (Synergy HT; Bio-Tek). The activities of β-lactamases for each strain were determined as A490 min−1 mg total protein−1.
Cells were viewed with a Zeiss Axiovert 200 microscope using differential interference contrast. Images were recorded with a Zeiss AxioCam MRc camera. To allow visualization of the cultures without any fixing and staining artifacts, 5 μl of cells was added to Delta T Dishes (Bioptechs Inc.) and pictures were taken from unprocessed samples obtained directly from the cultures.
The MspA structure was downloaded from the Protein Data Bank (identifier, 1uun). The two-dimensional structures of ampicillin (compound accession identifier [CID], 441087), chloramphenicol (CID, 5959), erythromycin (CID, 12560), kanamycin (CID, 6032), norfloxacin (CID, 4539), and vancomycin (CID, 441141) were downloaded from the PubChemCompound database at NCBI. Chem 3D Pro 8.0 software was used to obtain three-dimensional structures of the antibiotics and to minimize their energy using the built-in MOPAC algorithm. Finally, the antibiotic molecules were visualized using Chimera (41) (http://www.cgl.ucsf.edu/chimera). All molecules were drawn to scale and displayed as surface representations that showed both surface-accessible and solvent-excluded areas. Antibiotics were colored by atom. Molecular models of antibiotics were manually placed into the constriction zone of MspA.
The aim of this study was to examine which antibiotics utilize the porin pathway to enter mycobacteria. Since general porins of M. tuberculosis are still unknown, we used the model organism M. smegmatis, whose principal architecture of the outer membrane as the main permeability barrier is similar to that of slow-growing mycobacteria (21), to examine the physical properties of uptake pathways for several classes of antibiotics and TB drugs. The susceptibilities of two porin mutants, ML10 (ΔmspA ΔmspC) and ML16 (ΔmspA ΔmspC ΔmspD), to these agents were tested by Alamar blue assays and in agar dilution experiments (Table (Table1).1). Since the number of Msp porins in these mutants is at least 15- and 5-fold lower than in wt M. smegmatis and the ΔmspA mutant MN01, respectively (53), we expected that the mutants should show a more pronounced resistance phenotype if an antibiotic used the Msp pores for entry into M. smegmatis.
As a positive control, we determined the susceptibilities of ML10 and ML16 to hydrophilic β-lactam antibiotics, such as ampicillin and cephaloridine, which are known to diffuse through porins in gram-negative bacteria (39). The MICs of ampicillin for the porin mutants ML10 and ML16 were increased by 16- and 32-fold to 1,000 μg/ml in Alamar blue assays and in agar dilution experiments, respectively (Table (Table1;1; see Fig. S1 in the supplemental material). In comparison, deletion of the major porins OmpF and OmpC only moderately increased resistance to ampicillin to MICs of 8 and 16 μg/ml in E. coli (19, 23). The double and triple porin mutants ML10 and ML16 were also completely resistant to all other β-lactam antibiotics tested (Table (Table11).
Resistance to β-lactam antibiotics is determined by the synergy of the outer membrane permeability barrier and enzymatic hydrolysis by β-lactamases. M. smegmatis encodes several chromosomal β-lactamases. Alterations in their expression can drastically change the resistance of the strain, as was observed for TEM β-lactamase-producing E. coli and Salmonella enterica (6, 47). Therefore, we analyzed the activities of β-lactamases in the constructed porin mutants. M. smegmatis whole-cell extracts and culture filtrates were prepared as described earlier (30), and the activities of β-lactamases were determined using the chromogenic β-lactam nitrocefin. The vast majority of the β-lactamase activity of wt M. smegmatis was cell associated and was 20-fold higher than the activity of the culture filtrate (not shown), consistent with previously published results (30). Importantly, the β-lactamase activities of the porin mutants MN01, ML10, and ML16 did not differ significantly from that of wt M. smegmatis (not shown). Moreover, addition of 100 mg/ml ampicillin did not cause a significant increase in the β-lactamase activities of the wt and the porin mutant ML10 (see Fig. Fig.3).3). It was concluded that the increased resistance of the porin mutants to β-lactam antibiotics is not caused by faster hydrolysis of the drugs but is due to slower porin-mediated diffusion across the outer membrane. We used the Zimmermann-Rosselet assay (25, 60) to quantify the permeability of the outer membrane of M. smegmatis ML10 to cephaloridine. Hydrolysis of 0.8 mM cephaloridine was ninefold slower than that of wt M. smegmatis (not shown). Since the total β-lactamase activity in the supernatant and inside M. smegmatis cells did not change in the porin mutants, we concluded that deletion of the mspA and mspC genes significantly reduced the outer membrane permeability of ML10 to an extent similar to that shown previously for the ΔmspA mutant, MN01 (52). These results demonstrated a direct link between the porin-mediated outer membrane permeability and the susceptibility of M. smegmatis to cephaloridine and provided an immediate explanation for the increased resistance of M. smegmatis porin mutants to β-lactam antibiotics (Table (Table1).1). The observation that expression of only 15 to 30 MspA pores per μm2 of cell wall substantially increased the susceptibility of M. tuberculosis and M. bovis BCG to several β-lactam antibiotics (28) supports the conclusion that porins provide the major uptake pathway for β-lactams in mycobacteria.
Fluoroquinolones are very important second-line drugs for treatment of TB (32). However, it is unknown how fluoroquinolones cross the outer membranes of mycobacteria. In gram-negative bacteria, porins at least partially contribute to the uptake of fluoroquinolones (2, 4, 5). To assess a potential role of porins in this process in mycobacteria, we first determined the MICs of several fluoroquinolones for the wt strain and the porin mutants. M. smegmatis was very susceptible to all fluoroquinolones tested in this study, with MICs ranging from 0.1 to 3.2 μg/ml (Table (Table1).1). The MICs of ciprofloxacin, sparfloxacin, and moxifloxacin for M. smegmatis differed by not more than twofold from the MICs obtained earlier by agar dilution experiments (29, 45). The lowest MIC for M. smegmatis was measured for moxifloxacin, the most hydrophobic fluoroquinolone in this study (Table (Table1).1). The MIC of norfloxacin, the smallest and most hydrophilic fluoroquinolone, was 32-fold increased compared to that of moxifloxacin (Table (Table1;1; see Fig. S3 in the supplemental material). This is consistent with the previous observation that the activities of fluoroquinolones against mycobacteria increase with the hydrophobicity of the drugs (17) and may indicate a preference of the fluoroquinolones for direct diffusion through the lipid membranes of mycobacteria. On the other hand, increasing lipophilicity of the substituent at position N-1 did not correlate with increased potency against mycobacteria (46). We observed a fourfold increase in the resistance of the porin triple mutant ML16 to sparfloxacin and moxifloxacin compared to wt M. smegmatis (Table (Table1).1). To examine whether porins play a role in the uptake of fluoroquinolones, we employed an uptake assay based on the intrinsic fluorescence of fluoroquinolones. Therefore, we determined the absorption and fluorescence emission maxima of the four fluoroquinolones tested in this study. Only moxifloxacin and norfloxacin exhibited detectable fluorescence at 0.32 μg/ml when excited at the absorption maximum (not shown). Therefore, we chose both fluoroquinolones for the uptake assay. Figure Figure11 shows that the intracellular accumulation of norfloxacin in wt M. smegmatis is approximately four times higher than that in the porin mutant ML10, in contrast to the low adsorption of norfloxacin measured at 0°C (Fig. (Fig.1).1). Uptake of moxifloxacin could not be determined due to the high adsorption of the hydrophobic moxifloxacin to the cell surface of M. smegmatis (not shown). Taken together, these results show that porins play a role in the uptake of hydrophilic fluoroquinolones, such as norfloxacin, by M. smegmatis.
Deletion of Msp porins increased the resistance of M. smegmatis to chloramphenicol fourfold (Table (Table1),1), suggesting that the Msp porins contribute to the passage of chloramphenicol across the outer membrane of M. smegmatis. The uptake of [14C]chloramphenicol by both the wt and the porin mutant ML10 of M. smegmatis was assessed to examine the role of Msp porins in this process. Chloramphenicol was rapidly accumulated by wt M. smegmatis and reached a saturating concentration 2 minutes after addition of the antibiotic (Fig. (Fig.2).2). Importantly, accumulation of chloramphenicol was significantly slower in the ML10 strain, demonstrating that the Msp porins significantly contribute to the uptake of chloramphenicol by M. smegmatis. It should be noted that the concentration of chloramphenicol decreased after 2 minutes, suggesting that an efflux mechanism contributes to the resistance of M. smegmatis to the antibiotic (Fig. (Fig.2).2). A role of porins in the uptake of chloramphenicol and susceptibility was also demonstrated for E. coli (33) and Pseudomonas aeruginosa (22), suggesting that similar physical principles govern the passage of this antibiotic across the outer membranes of gram-negative bacteria and mycobacteria.
Deletion of the porin genes did not change the MICs of tetracycline, ethambutol, and isoniazid for M. smegmatis. The permeability mediated by the remaining porin, MspB, which differs from MspA by only two surface-exposed amino acids (13, 52), might still be sufficient for these small drugs to inactivate their target enzymes at similar rates in both M. smegmatis wt and the triple porin mutant, ML16. This might also be caused by the lack of another resistance mechanism, such as an efflux system, which is often required for significant resistance to be observed in porin mutants (55). Thus, the possibility that Msp porins contribute to the uptake of these drugs cannot be excluded.
In a previous study, it was reported that the MIC of vancomycin for the ΔmspA mutant MN01 was 10-fold increased compared to that of wt M. smegmatis (54). This was a surprising result, because vancomycin has molecular dimensions of 2.9 by 2.4 nm (49) and requires a channel at least 3.5 nm in diameter to cross the outer membrane of E. coli (26). However, the crystal structure of MspA revealed a channel constriction with a diameter of only 1 nm (13). Here, we examined the vancomycin susceptibilities of the porin mutant ML10, which lacks the two porin genes mspA and mspC that are expressed in wt M. smegmatis, and of the porin triple mutant ML16, which additionally lacks the mspD gene (53). Using the microplate Alamar blue assay and the agar dilution assay, we observed only twofold-increased resistance of the porin mutants ML10 and ML16 to vancomycin (Table (Table1).1). The increased MIC of the ΔmspA mutant, MN01, obtained in the previous study may have resulted from cell aggregates, which were observed more frequently for porin mutants (unpublished observations). The liquid cultures of M. smegmatis used in this study were free of aggregates larger than 5 to 10 cells as observed by microscopy (not shown). The MIC of kanamycin as a representative of aminoglycoside antibiotics was also only twofold increased for the porin double mutant ML10 (Table (Table1).1). Since uptake assays are not available for vancomycin and kanamycin, one cannot examine directly whether the Msp porins play any role in the transport of these antibiotics across the outer membrane of M. smegmatis. However, considering the large sizes of these antibiotics and the drastic reduction in the number of porins in the ML10 and ML16 mutants, a more significant MIC increase would be expected if these molecules diffused through the Msp porins. Taken together, these results suggest that Msp porins do not play a significant role in the uptake of vancomycin and aminoglycosides by M. smegmatis. This is similar to findings for E. coli porin mutants, whose susceptibilities to aminoglycosides and vancomycin were not altered (18). Since aminoglycosides and vancomycin do not appear to be affected by an impaired porin pathway and are too hydrophilic to diffuse through lipid membranes, other entry pathways, such as self-promoted uptake, as described for E. coli (18), need to be considered for mycobacteria as well.
The susceptibilities of E. coli and other gram-negative bacteria to hydrophobic antibiotics were not affected by deletion of porins (37). This was explained by the decreasing penetration rate through porins with increasing hydrophobicity of the solutes. The same physical principle also applies to mycobacterial porins. However, previous agar dilution experiments showed a two- to threefold-increased resistance to erythromycin and rifampin of the porin mutant MN01 compared to that of wt M. smegmatis (54). Using the Alamar blue assay, we reproducibly determined twofold-higher MICs of erythromycin and rifampin for all porin mutants compared to those of wt M. smegmatis (Table (Table1).1). Computer modeling indicated that erythromycin is too large to diffuse through the MspA pore (see Fig. Fig.4D).4D). This appears to exclude diffusion through Msp pores as the uptake pathway of these antibiotics in M. smegmatis. An alternative explanation is that the lack of porins in the outer membrane may indirectly decrease its permeability for large and hydrophobic solutes. Indeed, the M. smegmatis mspA mutant showed decreased permeability for the hydrophobic chenodeoxycholate (54). In conclusion, the efficacies of erythromycin and rifampin against M. smegmatis were not strongly affected by the absence of porins, consistent with the large size and the hydrophobicity of these antibiotics, which favor direct diffusion through the outer membrane lipid bilayer.
It should be noted that the previously determined MICs of erythromycin and rifampin for M. smegmatis were much higher than those in this study (54). We think that this was due to the formation of clumps and a concomitant decrease in accessibility of the cells in the previous study. However, this did not affect wt and porin mutants differently (54).
For most antibiotics tested in this study, the agar dilution method and the Alamar blue assay yielded the same or similar MICs for the different M. smegmatis strains (not shown). This is consistent with previous results obtained for many antibiotics for M. tuberculosis (7, 15). However, we noticed that the MICs of β-lactam antibiotics determined by the microplate Alamar blue assay were always higher than those determined by the agar dilution method (Table (Table11 shows ampicillin; data not shown for other β-lactam antibiotics). This is probably caused by the secretion of β-lactamases by M. smegmatis (Fig. (Fig.3),3), which decreases the effective concentrations of other β-lactams in the liquid medium in the Alamar blue assay more rapidly than in agar plates, resulting in higher MICs. Interestingly, higher MICs for β-lactam antibiotics in the Alamar blue assay were not observed for M. tuberculosis (7) and for M. bovis BCG (not shown). This may be due to lower β-lactamase activities of the slow-growing mycobacteria (14). We concluded that MICs for β-lactam antibiotics obtained by the Alamar blue assay should be interpreted with caution.
The size, hydrophobicity, and net charge are critical parameters determining the diffusion rate of a solute through a water-filled channel (37, 38). However, the primary function of porins is to convert the outer membrane into a molecular sieve. Thus, solutes too large to fit through the pores are excluded from passage. We therefore, set out to determine which of the antibiotics tested in this study would fit through the MspA channel. To this end, the antibiotic molecules were drawn to scale and displayed as surface representations that showed both surface-accessible and solvent-excluded areas (Fig. (Fig.4).4). The radius of the MspA constriction zone calculated by Chimera is about 1.6 nm. The second-largest dimensions of ampicillin, chloramphenicol, and norfloxacin when oriented along their longest axes are 0.5, 0.44, and 0.8 nm, respectively. These models clearly show that these antibiotics are significantly smaller than the constriction zone of MspA (Fig. 4A, B, and C) and corroborate experimental data that show that porins are used for uptake of these compounds (52) (Fig. (Fig.1).1). By contrast, the second-largest dimensions of erythromycin, kanamycin, and vancomycin when oriented along their longest axes are about 1.0 nm. It can be easily seen that these antibiotics either have the same dimensions as or are too large for the constriction zone of MspA. It has been shown by Nikaido and Rosenberg and Cowan et al. that the diffusion rate through porins is reduced drastically with increasing solute size and was close to zero for lactose, which has a hydrated radius only half of the size of the constriction zone of the main porin, OmpF, of E. coli (9, 40). Therefore, we concluded that erythromycin, kanamycin, and vancomycin are too large for diffusion through MspA. This is consistent with experiments in which wt and porin mutants of M. smegmatis showed similar susceptibilities to these antibiotics.
This is the first experimental evidence that hydrophilic fluoroquinolones and chloramphenicol diffuse through porins in mycobacteria. The finding that the hydrophilic norfloxacin uses mainly the Msp porins, while the hydrophobic moxifloxacin was the more effective antibiotic, indicates that direct diffusion through the outer membrane of M. smegmatis might be faster than diffusion through porins for the more hydrophobic fluoroquinolones. Thus, mutations resulting in less efficient porins or lower porin expression levels likely represent a mechanism for the opportunistic pathogens M. avium, M. chelonae, and M. fortuitum (59), which have Msp-like porins (36), to acquire resistance to fluoroquinolones. M. tuberculosis lacks Msp-like porins (35) but is nevertheless susceptible to both moxifloxacin and norfloxacin, which have MICs of 0.5 and 2 μg/ml, respectively (42). This may indicate that a porin-like protein of M. tuberculosis enables diffusion of norfloxacin across the outer membrane or that the slow growth of M. tuberculosis makes it susceptible to compounds with low outer membrane permeability. Obviously, more work is needed to distinguish between these possibilities and to understand the uptake of fluoroquinolones and of other TB drugs across the outer membrane of M. tuberculosis.
We thank Joachim Stephan and Jennifer Bender for initial experiments in this study and K. C. Walls for critically reading the manuscript.
This work was funded by grant AI63432 from the National Institutes of Health to M.N.
Published ahead of print on 16 June 2008.
†Supplemental material for this article may be found at http://aac.asm.org/.