Multidrug Resistance of a Porin Deletion Mutant of Mycobacterium smegmatis

Mycobacteria contain an outer membrane of unusually low permeabilitywhich contributes to their intrinsic resistance to many agents.It is assumed that small and hydrophilic antibiotics cross theouter membrane via porins, whereas hydrophobic antibiotics maydiffuse through the membrane directly. A mutant of Mycobacteriumsmegmatis lacking the major porin MspA was used to examine therole of the porin pathway in antibiotic sensitivity. Deletionof the mspA gene caused high-level resistance of M. smegmatisto 256 µg of ampicillin/ml by increasing the MIC 16-fold.The permeation of cephaloridine in the mspA mutant was reducedninefold, and the resistance increased eightfold. This establisheda clear relationship between the activity and the outer membranepermeation of cephaloridine. Surprisingly, the MICs of the largeand/or hydrophobic antibiotics vancomycin, erythromycin, andrifampin for the mspA mutant were increased 2- to 10-fold. Thisis in contrast to those for Escherichia coli, whose sensitivityto these agents was not affected by deletion of porin genes.Uptake of the very hydrophobic steroid chenodeoxycholate bythe mspA mutant was retarded threefold, which supports the hypothesisthat loss of MspA indirectly reduces the permeability by thelipid pathway. The multidrug resistance of the mspA mutant highlightsthe prominent role of outer membrane permeability for the sensitivityof M. smegmatis to antibiotics. An understanding of the pathwaysacross the outer membrane is essential to the successful designof chemotherapeutic agents with activities against mycobacteria.

INTRODUCTION

Top

Introduction

The prevalence and spread of antibiotic resistance are increasinglyserious problems that hamper the effective treatment of infectiousdiseases (26). The search for new antibiotics is mainly basedon novel bacterial targets and high-throughput screening assays(10). However, many lead compounds discovered in vitro may failbecause they do not reach their targets at sufficiently highconcentrations in vivo (7). This is true in particular for gram-negativebacteria, which, in contrast to gram-positive bacteria, areprotected from the toxic actions of certain antibiotics, dyes,and detergents and host defense factors such as lysozyme byan additional outer membrane (OM) (49). The OM can be crossedby at least two general pathways: the hydrophobic (or lipid)pathway, which is characterized by the nature and the interactionsof the membrane lipids, and the hydrophilic (or porin) pathway,whose properties are determined by water-filled channel proteins,the porins, which span the OM of gram-negative bacteria (49).It has been shown by the pioneering work of Nikaido and collaborators(21, 46) that Escherichia coli and Salmonella porins play amajor role in the transport of ß-lactam antibiotics.Subsequent studies showed that porin-deficient mutants of gram-negativebacteria were also more resistant to quinolones, tetracyclines,chloramphenicol, nalidixic acid, and trimethoprim (6, 15, 25,52). These data suggest that porins are involved in the transportof these antibiotics, but the levels of resistance were ratherlow, as reflected by the 1.5- to 4-fold increased MICs for theporin mutants. However, combination of low-level OM permeabilitywith drug efflux produces high-level resistance in gram-negativebacteria (44, 45) and is a powerful mechanism leading to multidrugresistance (51).

The permeability of the cell envelope of mycobacteria is severalorders of magnitude lower than that of the cell envelope ofgram-negative bacteria, which is believed to be the major reasonfor the intrinsic resistance of mycobacteria to most antibioticsand chemotherapeutic agents (31). This is likely due to an uniquearrangement of the longest fatty acids known to date in nature,the mycolic acids, which can consist of up to 90 carbon atomsand which are covalently bound to a single polymeric head group,the arabinogalactan (4). Porins in the mycobacterial OM mediatethe diffusion of hydrophilic solutes, although the efficiencyof this pathway is 100- to 1,000-fold lower than that in E.coli (32). The porin MspA was identified in CHCl₃-methanol andin detergent extracts of Mycobacterium smegmatis (24, 41). Expressionof the mspA gene encoding the mature MspA protein in E. coliyielded a folded but inactive 20-kDa monomer (23) and an oligomerwith an apparent molecular mass of 100 kDa, which formed channelssimilar to those of the native protein (41). The genome of M.smegmatis encodes three additional very similar porins, designatedMspB, MspC, and MspD, that differ from MspA only at 2, 4, and18 positions of the mature protein, respectively (61). Detergentextracts of a ${Delta}$ mspA M. smegmatis mutant exhibited significantlylower channel activities in lipid bilayer experiments and containedfewer porins than extracts of wild-type M. smegmatis. The lowerporin content reduced the cell wall permeability of the ${Delta}$ mspAmutant for cephaloridine and glucose nine- and fourfold, respectively.However, the growth rate was not significantly altered upondeletion of the mspA gene (61). These results show that MspAis the main general diffusion pathway for hydrophilic moleculesacross the OM of M. smegmatis and is only partially replacedby fewer porins in the cell wall of the ${Delta}$ mspA mutant. Other fast-growingmycobacteria, such as M. chelonae and M. phlei, appear to encodeporins similar to MspA (41, 54), but no sequence homologs wereidentified in slowly growing mycobacteria, such as M. tuberculosis,or in any other organism outside the genus Mycobacterium (40).Determinants of the low efficiency of the porin pathway in M.smegmatis are the 45-fold lower number of pores and the 2.5-foldlonger pore channels compared to the numbers and the sizes ofthe pores in E. coli (12). Similar causes for low permeabilityare likely to exist for all mycobacteria (40). However, therelative importance of the porin pathway and the lipid pathwayacross the OM of mycobacteria is unknown for any antibiotic,although the synergistic effects of cell wall-damaging agentssuch as ethambutol with many other drugs have been known fora long time (22, 67).

To examine the role of porins in the antibiotic sensitivityof M. smegmatis, we determined the MICs of many antibioticsfor an M. smegmatis mspA mutant. Furthermore, we did transportexperiments using radiolabeled compounds to confirm that reducedsensitivity correlated with slower uptake. This study demonstratesfor the first time the importance of porins for the sensitivityof M. smegmatis to both hydrophilic and hydrophobic antibiotics.

MATERIALS AND METHODS

Top

Materials and Methods

Chemicals, bacterial strains, media, and growth conditions. Chemicals were purchased from Merck, Roth, or Sigma at the highestpurity available. [¹⁴C]chenodeoxycholate was obtained from NENLife Science Products, and [¹⁴C]isoniazid was obtained fromHartmann Analytic. M. smegmatis SMR5 is a derivative of M. smegmatismc²155 and is streptomycin resistant due to a mutation in ribosomalprotein S12 (55). The construction of ${Delta}$ mspA mutant M. smegmatisMN01 was described previously (61). All mycobacterial strainswere grown in Middlebrook 7H9 liquid medium (Difco) supplementedwith 0.2% glycerol and 0.05% Tween 80 or on Middlebrook 7H10plates (Difco) containing 0.2% glycerol at 37°C.

Drug sensitivity by agar dilution. The MICs for M. smegmatis SMR5 and M. smegmatis MN01 were determinedin triplicate by serial dilutions in 7H10 agar. Each strainwas first grown in a 4-ml culture for 2 days at 37°C toan optical density at 600 nm of 0.6 to 0.8. The cultures werediluted with 7H9 liquid medium to yield approximately 5,000CFU per ml. Approximately 500 CFU was streaked on plates containingthe appropriate antibiotic concentrations. As a reference, 500CFU was also streaked on plates without antibiotic. The numberof surviving cells was normalized to the number of cells countedon plates without antibiotic for each strain and was expressedas the relative CFU (percent CFU). For all drugs tested, theconcentration that resulted in the complete inhibition of bacterialgrowth after 3 days of incubation at 37°C was determinedas the MIC of the drug.

Transport experiments. Transport experiments were carried out as described previously(61). To reduce aggregation and clumping of the cells, M. smegmatisSMR5 and MN01 were each first grown in 4-ml cultures for 2 daysat 37°C and then filtered through a 5-µm-pore-sizefilter (Sartorius). The filtrates were grown for 2 days at 37°Cand then used to inoculate 100-ml cultures, which were grownto an optical density at 600 nm of 0.5. The cells were harvestedby centrifugation (1,250 x g at 4°C for 10 min), washedonce in 2 mM piperazine-N,N'-bis(2-ethanesulfonic acid) (pH6.5)-0.05 mM MgCl₂, and resuspended in the same buffer. The¹⁴C-labeled compounds and their nonlabeled analogs were mixedand added to the cell suspensions to obtain final concentrationsof 5 and 20 µM. The mixtures were incubated at 37°C,and 1-ml samples were removed at the indicated times. The cellswere filtered through a 0.45-µm-pore-size filter (Sartorius),washed with 0.1 M LiCl, and counted in a liquid scintillationcounter. The mean dry weight of the cells in these samples was1.4 ± 0.4 mg. The uptake rate was expressed as nanomolesper milligram of cells.

Analysis of cell wall lipids of M. smegmatis. M. smegmatis mc²155 and its isogenic porin-deficient mutantstrain, MN01, were grown as surface pellicles on Sauton's mediumfor 6 days at 37°C. The strains were maintained by selectionwith streptomycin, which was added to a final concentrationof 20 µg/ml. Lipids were isolated and characterized asdescribed previously (13). Wet cells were first extracted withchloroform-methanol (1:2; vol/vol), then five times with chloroform-methanol(1:1; vol/vol), and finally, once with chloroform-methanol (2:1;vol/vol). These extracts were pooled, dried under vacuum, andpartitioned between water and chloroform (1:1; vol/vol). Theorganic phase was extensively washed with distilled water, evaporatedto dryness to yield the crude lipid extract, and weighed. Cell-boundfatty acids (mycolates) were obtained by saponification of delipidatedcells with 5% KOH in methanol-benzene (8:2; vol/vol) for 16h at 80°C. After acidification with sulfuric acid, fattyacids were extracted with diethyl ether and extensively washedwith water. The washed lipid extract was dissolved in chloroformand analyzed by thin-layer chromatography (TLC) on silica gelplates (thickness, 0.25 mm; Macheray-Nagel) precoated with Durasil25. Lipid spots were resolved by TLC run in the following solventmixtures: petroleum ether-diethyl ether (9:1; vol/vol) for analysisof triacyl glycerols (TAGs), chloroform-methanol (9:1; vol/vol)for analysis of glycopeptidolipids (GPLs) and trehalose dimycolates(TDMs), and chloroform-methanol-water (60:30:8; vol/vol) foranalysis of trehalose monomycolates (TMMs) and phospholipids.Sugar-containing compounds (GPLs, TDMs, TMMs and phosphatidylinositolmannosides) were visualized by spraying the plates with 0.2%anthrone in concentrated sulfuric acid, followed by heatingat 110°C. The Dittmer-Lester reagent (9) was used to detectphosphorus-containing substances. The ninhydrin reagent wasused to reveal the presence of free amino groups. A spray with10% molybdophosphoric acid in ethanol solution, followed byheating at 110°C, was used to detect all of the lipids spots,including TAGs.

RESULTS

Top

Results

The mspA mutant of M. smegmatis is more resistant to most ß-lactam antibiotics. It has been well established that ß-lactam antibioticscross the OM of gram-negative bacteria by diffusing throughporin channels (29, 48, 56). Therefore, we first examined whetherdeletion of the major porin MspA of M. smegmatis has any influenceon the resistance of mycobacteria to ß-lactam antibiotics.MICs were determined by agar dilution with Middlebrook 7H10plates. Figure 1A shows that wild-type strain M. smegmatis SMR5is drastically more susceptible to the zwitterionic agent ampicillinthan the isogenic mspA mutant strain, M. smegmatis MN01. Thus,deletion of mspA increased the ampicillin MIC for M. smegmatisby 16-fold, to 256 µg/ml (Table 1). Expression of themspA gene on a plasmid (pMN014) in M. smegmatis MN01 establishedagain the sensitivity of the wild type, demonstrating that theloss of mspA caused the resistance phenotype. M. smegmatis wasalso more sensitive to the zwitterionic agent cephaloridinethan to ampicillin. Deletion of the mspA gene increased thecephaloridine MIC fourfold (Table 1). No significant differencein the susceptibilities of the two strains to the anionic agentcephalothin was observed (Table 1). It should be noted thatthe growth rates of both strains are similar in standard 7H9and minimal liquid media (61) and on 7H10 agar plates.

View larger version (17K):
[in this window]
[in a new window]
FIG. 1. MspA-dependent sensitivity of M. smegmatis to hydrophilic antibiotics. The sensitivities of wild-type M. smegmatis (black bars), the ${Delta}$ mspA mutant (white bars), and the ${Delta}$ mspA mutant transformed with the mspA expression vector pMN014 (gray bars) to ampicillin (A) and vancomycin (B) were determined by serial dilution in Middlebrook 7H10 agar. The number of surviving cells was normalized to the number cells counted on plates without antibiotic for each strain, and the results are expressed as relative CFU (percent CFU). The sensitivity experiments were done in triplicate, and the concentrations (c) are shown with their standard deviations. The structures of the antibiotics are shown in the corresponding graphs. The asterisks denote the datum points that differed significantly (P < 0.05) between the wild-type strain and the mspA mutant strain by the paired Student t test.

View this table:
[in this window]
[in a new window]
TABLE 1. Physicochemical properties of antibiotics and the efficiencies of their activities against M. smegmatis^a

The susceptibility of M. smegmatis to vancomycin but not to aminoglycosides is affected by deletion of the mspA gene. Hydrophilic solutes up to an exclusion limit of approximately600 to 700 Da are thought to prefer the porin pathway acrossthe OM of gram-negative bacteria (43). Since the exclusion limitof mycobacterial porins for hydrophilic solutes is unknown,we examined whether the susceptibility of M. smegmatis to hydrophilicantibiotics larger than ß-lactam antibiotics was alsoaffected by deletion of the mspA gene. The susceptibility ofM. smegmatis to kanamycin was high, and that of the mspA mutantwas increased only slightly (data not shown). This result issimilar to that obtained for porin mutants of E. coli, whichdid not show altered susceptibilities to aminoglycoside antibiotics(16, 18). Surprisingly, the mspA mutant was clearly more resistantto the much larger hydrophilic peptide antibiotic vancomycin(Fig. 1B). This result was also reflected by a 10-fold increasedMIC (Table 1). It should be noted that only a small populationof vancomycin-resistant cells caused the increased MIC for themspA mutant (Fig. 1B) and that M. smegmatis was 50-fold moresensitive to vancomycin than E. coli (64).

The mspA mutant of M. smegmatis is more resistant to hydrophobic antibiotics. The rates of penetration of compounds through water-filled proteinchannels such as the porins decrease with increasing hydrophobicity(46). This explains why E. coli mutants lacking one or moreporins have nearly unaltered sensitivities to hydrophobic antibiotics(49). A similar behavior was expected for porin mutants of M.smegmatis. Surprisingly, deletion of the mspA gene increasedthe resistance of M. smegmatis to the hydrophobic antibioticerythromycin, resulting in a more than twofold increased MIC(Fig. 2A; Table 1). The level of resistance mediated by theloss of MspA was even more pronounced than that obtained forthe hydrophilic antibiotic cephaloridine, although this is notreflected in the MICs of the two antibiotics (Table 1). Thesusceptibility of the mutant to erythromycin was restored byexpression of the mspA gene (Fig. 2A), demonstrating that deletionof the mspA gene caused this resistance phenotype. The sensitivitiesof M. smegmatis MN01 to rifampin, novobiocin, and chloramphenicolwere determined to examine whether deletion of mspA also affectedsensitivities to other hydrophobic compounds or was an erythromycin-specificeffect. Rifampin was less effective against M. smegmatis MN01than against the wild type (Table 1). Similarly, the hydrophobicantibiotics novobiocin and chloramphenicol were less effectiveagainst the mspA mutant, although the difference was smallerthat that for erythromycin and was detectable at only a fewconcentrations (data not shown). These results indicate thatthe reduced porin expression increased the level of resistanceof the mspA mutant not only to hydrophilic antibiotics but alsoto hydrophobic antibiotics.

View larger version (16K):
[in this window]
[in a new window]
FIG. 2. MspA-dependent sensitivities and permeabilities of M. smegmatis to hydrophobic compounds. (A) The sensitivities of wild-type M. smegmatis (black bars), the ${Delta}$ mspA mutant (white bars), and the ${Delta}$ mspA mutant transformed with the mspA expression vector pMN014 (gray bars) to erythromycin were determined by serial dilution in Middlebrook 7H10 agar (c, concentration). The number of surviving cells was normalized to the number of cells counted on plates without antibiotic for each strain, and the results are expressed as relative CFU (percent CFU). (B) The accumulation of [¹⁴C]chenodeoxycholate by M. smegmatis SMR5 (wild type; closed circles) and M. smegmatis MN01 ( ${Delta}$ mspA; open circles) was measured. Regression analysis yielded uptake rates of 0.09 and 0.03 nmol mg^–1 min^–1 for 20 µM chenodeoxycholate for the wild-type and mutant strains of M. smegmatis, respectively. The assay was performed at 37°C. The structures of the compounds are shown in the corresponding graphs. The sensitivity and uptake experiments were done in triplicate, and the results are shown with their standard deviations. The asterisks denote the datum points that differed significantly (P < 0.05) between the wild-type strain and the mspA mutant strain by the paired Student t test.

Considering the reduced susceptibility of the M. smegmatis mspAmutant to hydrophobic antibiotics, we suspected that a lessefficient permeation of hydrophobic solutes might be responsiblefor this phenotype. To this end, we measured the rates of uptakeof the very hydrophobic steroid chenodeoxycholate, which isconsidered an indicator of the permeability of the lipid pathwayin mycobacteria, by M. smegmatis and the mspA mutant (28, 36).The uptake rate was 0.09 nmol · min^–1 ·mg^–1 for the wild type, but the rate was reduced threefoldfor the mspA mutant (Fig. 2B). This result indicates that, indeed,a slower rate of transport of hydrophobic compounds across theOM increased the level of resistance of the mspA mutant to hydrophobicantibiotics.

Quantitative analysis of the cell wall lipids of the M. smegmatis wild type and the ${Delta}$ mspA strain. Alteration of the lipid composition of the cell wall of M. smegmatismight explain the decreased permeability of the ${Delta}$ mspA mutantto hydrophobic compounds. To test this possibility, noncovalentlybound lipids of the cell wall of the M. smegmatis wild typeand the ${Delta}$ mspA mutant were extracted with organic solvents andanalyzed by TLC. The cell-bound fatty acids (mycolates) wereobtained by saponification of the delipidated cells and extractionwith ether oxide. The following lipids were detected in similaramounts in both strains of M. smegmatis: TAGs, GPLs, TDM, TMM,phosphatidylinositol mannosides 2 and 6, phosphatidylinositol,phosphatidylglycerol, and phosphatidylethanolamine. These resultsshow that deletion of the porin gene mspA does not significantlyalter the lipid composition of the M. smegmatis cell wall.

Deletion of the mspA gene does not affect the susceptibility of M. smegmatis to small and hydrophilic chemotherapeutic agents for tuberculosis. Mycolic acids and arabinogalactan are essential components ofthe mycobacterial cell wall. Isoniazid and ethambutol interferewith the synthesis of mycolic acids (65) and with the assemblyand synthesis of the arabinogalactan (62), respectively. Bothdrugs are essential in present tuberculosis chemotherapy regimens(34). Since isoniazid and ethambutol are small and hydrophilicmolecules, it was assumed that they use the porin pathway forentry into mycobacteria (35). Similar considerations apply tocycloserine, which inhibits peptidoglycan synthesis and whichis used as a second-line drug in tuberculosis chemotherapy.Despite these assumptions, the effect of deletion of the mspAgene on the sensitivity of M. smegmatis to these drugs was negligible.The mspA mutant showed marginally increased levels of resistanceto all three drugs only at high concentrations. This is shownfor isoniazid in Fig. 3A . Interpretation of the results ofdrug sensitivity experiments in terms of cell wall permeabilityis always difficult and might be additionally hampered for isoniazid,because M. smegmatis is naturally about 100-fold more resistantto isoniazid than M. tuberculosis (66). Therefore, we examinedthe transport of 20 µM [¹⁴C]isoniazid directly. The amountof isoniazid associated with wild-type M. smegmatis was significantlyhigher than that associated with the mspA mutant, but it didnot increase between 1 and 15 min for both strains in severalindependent experiments at 37°C (Fig. 3B), indicating thatan equilibrium was already reached in these experiments. Toslow the uptake rate, the concentration of isoniazid was reducedto 5 µM and the experiments were performed at 20°C.Again, more isoniazid was associated with wild-type M. smegmatisthan with the mspA mutant, but no significant uptake was observedat between 15 and 90 s after addition of isoniazid to the cellsuspensions (data not shown).

View larger version (13K):
[in this window]
[in a new window]
FIG. 3. MspA-dependent sensitivity and permeability of M. smegmatis to isoniazid. (A) The sensitivities of wild-type M. smegmatis (black bars), the ${Delta}$ mspA mutant (white bars), and the ${Delta}$ mspA mutant transformed with the mspA expression vector pMN014 (gray bars) to isoniazid were determined by serial dilution in Middlebrook 7H10 agar (c, concentration). The number of surviving cells was normalized to the number of cells counted on plates without antibiotic for each strain, and the results are expressed as relative CFU (percent CFU). The structure of isoniazid is shown. (B) The accumulation of 20 µM [¹⁴C]isoniazid by M. smegmatis SMR5 (wild type; closed circles) and M. smegmatis MN01 ( ${Delta}$ mspA; open circles) was measured. The assay was performed at 37°C. The sensitivity and uptake experiments were done in triplicate, and the results are shown with their standard deviations. The asterisks denote the datum points that differed significantly (P < 0.05) between the wild-type strain and the mspA mutant strain by the paired Student t test.

DISCUSSION

Top

Discussion

The intrinsic resistance of mycobacteria to most antibioticsand chemotherapeutic agents is believed to be a result of theirimpermeable OM in synergy with other resistance mechanisms,such as enzymatic inactivation or active efflux of the drugs(4, 31). The role of the porin pathway across the OM in thesensitivity of M. smegmatis to antibiotics was examined by usinga mutant of M. smegmatis lacking the main porin MspA. The MICof cephaloridine for this mutant increased eightfold, and theOM of the mutant had a ninefold reduced permeability to cephaloridine,while the ß-lactamase activity of M. smegmatis wasnot altered (61). Thus, slower uptake of cephaloridine is directlycorrelated with increased resistance. These results indicatefurther that cephaloridine mainly diffuses through the MspAchannel across the OM of M. smegmatis. Similar conclusions weredrawn for E. coli, in which the outstanding permeation of cephaloridinethrough the porin OmpF was paralleled by its strong activity(46). However, it should be noted that there are two major differencesbetween M. smegmatis and E. coli. First, M. smegmatis is abouteight-fold less sensitive to cephaloridine than E. coli (MIC= 2 µg/ml) (21), consistent with a less permeable OM.Second, deletion of the porin gene mspA clearly increased thelevel of resistance of M. smegmatis to cephaloridine in contrastto that of E. coli, for which inactivation of the OmpF porindid not detectably alter the cephaloridine MIC (21, 29). Thecephaloridine MIC for the mspA mutant was even higher than thatfor an E. coli mutant lacking both major porins, OmpC and OmpF(29). Deletion of mspA produced high-level resistance of M.smegmatis to ampicillin by increasing the MIC 16-fold (MIC =256 µg/ml). This is the largest increase in the MIC ofan antibiotic ever observed for a single-porin-knockout strain,in which no or only small effects of gene knockouts on drugsensitivity are usually shown (21). Such a high level of resistanceto ampicillin was not observed even when the two main porinsof E. coli, OmpF and OmpC, were inactivated (29), underliningthe importance of MspA for sensitivity to ampicillin. M. smegmatiswas resistant to a 15-fold higher concentration of the monoanionicagent cephalothin (Table 1), consistent with the 10-fold lowerOM permeability of cephalothin compared to that of the zwitterionicagent cephaloridine (63). The cephalothin MIC for the mspA mutantwas increased only slightly, indicating that negative chargesseverely hamper diffusion through MspA. This is explained bythe crystal structure of MspA, which shows two rings, each ofwhich has eight aspartate residues, lining the channel at theconstriction zone (14). Thus, penetration of cephalothin througha nonporin pathway may be significant, consistent with observationsfor other ß-lactam antibiotics, which have low ratesof diffusion through E. coli porins (47). Taken together, theseresults indicate that hydrophilic and cationic or zwitterionicß-lactam antibiotics cross the OM of M. smegmatisvia the porin pathway, similar to the situation in E. coli,in which most ß-lactams were found to diffuse throughporins (46, 47).

Isoniazid and ethambutol are small and hydrophilic drugs indispensablein present tuberculosis chemotherapy regimens (34) and are thoughtto penetrate the mycobacterial cell via porins (35). The sameholds true for the second-line antituberculosis drug cycloserine.We observed an increased level of resistance of the mspA mutantto these drugs only at concentrations close to the MIC (Fig.3A). Since small and hydrophilic molecules diffuse through porinsvery fast, permeation through the remaining porins of the mspAmutant, MspB, MspC, and MspD, might still be sufficient forthese drugs to inactivate their target enzymes at similar ratesin both strains. However, we cannot exclude the possibilitythat MspA does not contribute to the uptake of these drugs.We also could not measure any isoniazid uptake for M. smegmatisbetween 15 s and 15 min, in contrast to the uptake results inexperiments with M. tuberculosis (3). Whether this is due toa faster drug efflux in M. smegmatis or to other mechanismsis not known.

The kanamycin MIC for the mspA mutant of M. smegmatis was notreduced. This is similar to the findings for E. coli porin mutants(18). These results could be explained by a very efficient diffusionof aminoglycosides through the remaining porins in the mutants,as suggested earlier for E. coli on the basis of liposome swellingexperiments (39). An alternative route of cell entry for aminoglycosideswas proposed for Pseudomonas aeruginosa, which has an exceptionallylow permeability by the porin pathway (17). Aminoglycosideswere shown to permeabilize the OM of P. aeruginosa (19) andE. coli (18) and thereby promote their own uptake. Self-promoteduptake may be significant for M. smegmatis when the even lowerporin-mediated OM permeability in M. smegmatis compared to thatin P. aeruginosa is considered (30) and might explain the unchangedsensitivity of the mspA mutant of M. smegmatis to kanamycin.

Surprisingly, the vancomycin MIC for the mspA mutant was increased10-fold. It has been shown that OM proteins with large channels,such as TolC, which has a channel diameter of 3.5 nm and whichtranslocates folded polypeptides across the OM of E. coli (33),or the phage export channel protein pIV (38), are needed topromote vancomycin uptake in E. coli. However, the crystal structureof MspA revealed a channel constriction with a diameter of 1nm (14), which is too small to allow passage of vancomycin,which has molecular dimensions of 2.9 by 2.4 nm (57). Thus,it appears very unlikely that vancomycin can diffuse throughthe MspA channel. It should be noted that only a few percentof wild-type and mspA mutant cells were resistant to vancomycinat concentrations of 0.5 µg/ml or higher, indicating thatthe majority of the cells of both strains were killed at concentrationsfar below the MIC (Fig. 1B). This is reminiscent to the observationof "heteroresistant staphylococci," in which the majority ofa Staphylococcus aureus population is vancomycin susceptibleand a minority population is vancomycin resistant (60). Deletionof mspA apparently increased the MIC for the resistant subpopulationof M. smegmatis. A thickened cell wall has been shown to beassociated with vancomycin resistance in staphylococci (8).In the genus Enterococcus, resistance to vancomycin relies onthe exchange of the terminal dipeptide in pentapeptide cellwall precursors (2). However, the molecular mechanisms of vancomycinresistance in mycobacteria are unknown.

The hyperresistance of the mspA mutant of M. smegmatis to thelarge hydrophobic antibiotics rifampin and erythromycin wasunexpected for two reasons. First, the rates of diffusion ofß-lactam antibiotics through the porins of E. coliinversely depend on their hydrophobicities and are correlatedwith their efficacies (48). This was consistent with the observationsthat the sensitivities of E. coli and other gram-negative bacteriato hydrophobic antibiotics were not affected by the deletionof porins (49, 56) but were increased by mutation of the lipopolysaccharideor damage to the OM (42, 49). The highly negatively chargedconstriction site of the MspA channel (14) likely creates shellsof strongly bound water molecules which should exclude hydrophobicsolutes from passage. A similar mechanism, albeit based on avery different structure, was postulated and confirmed for theporin OmpF of E. coli (27, 58). Second, in addition to beinghydrophobic, the large sizes of erythromycin and rifampin makeit rather unlikely that these antibiotics diffuse through theMspA pore at a significant rate. An indirect effect of MspAon the OM permeability of M. smegmatis provides an alternativeexplanation for the increased resistance of the mspA mutantM. smegmatis to hydrophobic antibiotics. Alteration of the permeabilityof the lipid pathway by deletion of MspA is supported by thefinding that the rate of uptake of the very hydrophobic agentchenodeoxycholate by the mspA mutant is reduced threefold comparedto that by the wild type. Decreased permeability to chenodeoxycholatewas also observed in a mutant of M. tuberculosis lacking oxygenatedmycolic acids (11), which shows that the lipid arrangement isimportant for the permeability of the mycobacterial OM. Heterologousexpression of a very small number of MspA porins increased thepermeability of M. tuberculosis for the nucleic acid-binding,membrane-permeant stain SYTO 9 by 10-fold, indicating that MspAhas a significant influence on OM properties (37). Thus, thelipid arrangement in the mycobacterial OM appears to dependon both the mycolic acid and the protein compositions. Althoughthis model would explain the unexpected cross-resistance ofthe mspA mutant of M. smegmatis to hydrophobic antibiotics,it is in contrast to observations for Salmonella, in which reducedlevels of OM proteins increased the rates of penetration ofhydrophobic compounds (1). Experiments probing the fluidityof the OM of porin mutants are needed to solve this apparentcontradiction. Taken together, this study highlights the importanceof porins for the antibiotic resistance of M. smegmatis. Thisis of direct medical relevance for the treatment of nosocomialinfections caused by the opportunistic pathogens M. chelonaeand M. fortuitum, which have Msp-like porins (41), which areprevalent in up to 90% of biofilms (59), and which are associatedwith wound and catheter infections (5). Since the principalstructure of the cell wall appears to be similar for many mycobacteria(50), it is anticipated that porins also play a role in determiningthe drug sensitivity of M. tuberculosis. However, the majorityof clinically relevant multidrug-resistant M. tuberculosis strainsarise as a result of the sequential accumulation of resistancemutations in known genes, which either encode the targets orare involved in activation of the individual drugs. Nevertheless,M. tuberculosis strains with unknown resistance mechanisms existfor each of the primary antituberculosis drugs (53), and suchstrains should be analyzed for mutations involving porin genesand/or drug efflux.

ACKNOWLEDGMENTS

We thank Susanne Kubetzko for initial help with sensitivityexperiments and Wolfgang Hillen for continuous support.

This work was funded by the Deutsche Forschungsgemeinschaft(grant NI 412) and the European Union (5th Framework Programme,contract QLK2-2000-01761).

FOOTNOTES

* Corresponding author. Present address: Department of Microbiology, University of Alabama at Birmingham, 609 Bevill Biomedical Research Building, 845 19th St. South, Birmingham, AL 35294. Phone: (205) 996-2711. Fax: (205) 934-9256. E-mail: mnieder{at}uab.edu.

REFERENCES

Top