MspA is an extremely stable, oligomeric porin from Mycobacterium smegmatis that forms water-filled channels in vitro. Immunogold electron microscopy and an enzyme-linked immunosorbent assay demonstrated that MspA is localized in the cell wall. An mspA deletion mutant did not synthesize detectable amounts of mspA mRNA, as revealed by amplification using mspA-specific primers and reverse-transcribed RNA. Detergent extracts of the ΔmspA mutant exhibited a significantly lower porin activity in lipid bilayer experiments and contained about fourfold less porin than extracts of wild-type M. smegmatis. The chromosome of M. smegmatis encodes three proteins very similar to MspA. Sequence analysis of the purified porin revealed that mspB or mspC or both genes are expressed in the ΔmspA mutant. The properties of this porin, such as single channel conductance, extreme stability against denaturation, molecular mass and composition of 20 kDa subunits, are identical to those of MspA. Deletion of mspA reduced the cell wall permeability towards cephaloridine and glucose nine- and fourfold respectively. These results show that MspA is the main general diffusion pathway for hydrophilic molecules in M. smegmatis and was only partially replaced by fewer porins in the cell wall of the ΔmspA mutant. The minimal permeability coefficient of the ΔmspA mutant for glucose was 7.2 × 10−8 cm s−1, which is the lowest value reported so far for bacteria. This is the first experimental evidence that porins are the major determinants of the exceptionally low permeability of mycobacteria to hydrophilic molecules.
Mycobacterium tuberculosis devotes a large proportion of the coding capacity of its chromosome to lipid biosynthesis (Cole et al., 1998). Most of its lipids are part of an unique cell wall, the composition and structure of which have been described in detail (Brennan and Nikaido, 1995; Draper, 1998; Dmitriev et al., 2000). The most prominent functional feature of the mycobacterial cell wall is its very low permeability, which protects mycobacteria from noxious substances. Diffusion of hydrophilic molecules across the mycolic acid layer occurs via porins (Trias et al., 1992). Although the low fluidity of the mycolic acid layer directly explains its function as a barrier against hydrophobic molecules (Liu et al., 1996), it is not understood why the hydrophilic pathway of mycobacteria is 100- to 1000-fold less efficient than that of Escherichia coli (Jarlier and Nikaido, 1990).
We have identified MspA as the first porin of M. smegmatis (Niederweis et al., 1999). MspA is an extremely stable oligomeric protein exhibiting a large channel conductance of 4.6 nS in 1 M potassium chloride. A significant fraction of isolated MspA forms channels after extraction with organic solvents, after treatment with 8 M urea or after boiling in 3% SDS. This stability was exploited for selective extraction of MspA from M. smegmatis cells at high temperatures (Heinz and Niederweis, 2000). It was concluded that MspA is the prototype of a new family of channel-forming proteins, because several DNA fragments homologous to the mspA gene were observed in the chromosome of M. smegmatis (Niederweis et al., 1999). However, no other member of this protein family has been identified yet. In addition, nothing is known about the structure of MspA or its physiological function.
The aim of this study was to demonstrate the cell wall localization of MspA and to analyse whether MspA is a significant part of the hydrophilic pathway through the cell wall of M. smegmatis. We showed that an mspA deletion mutant expressed another porin with properties very similar to those of MspA. The permeability of the cell wall of the ΔmspA mutant towards cephaloridine was reduced ninefold compared with wild-type M. smegmatis, and the transport of glucose was impaired. This study demonstrates that MspA plays an important role in the cell wall permeability of M. smegmatis and is the first proof of the physiological function of a porin from Gram-positive bacteria.
Results and discussion
Construction of an mspA deletion mutant of M. smegmatis
To analyse the function of the MspA porin in M. smegmatis, we wanted to delete the mspA gene rather than inactivate it in order to avoid instability problems as a result of reversion events, which were often observed in E. coli strains lacking the porins OmpC and OmpF (Nogami and Mizushima, 1983). For this purpose, we constructed a ΔmspA::aacC1 cassette that conferred resistance against gentamicin and was flanked by 1400 bp and 600 bp of chromosomal DNA located upstream and downstream of the mspA gene respectively (Fig. 1A). In this cassette, 536 bp containing the putative promoter, ribosome binding site and 40% of the mspA gene encoding the signal peptide and 54 amino acids of the mature protein were deleted. To improve the low frequency of homologous recombination in mycobacteria (McFadden, 1996), the ΔmspA::aacC1 cassette was used in the plasmid pMN226inv, which carries two counterselectable markers, the rpsL gene (Sander et al., 1995) and a temperature-sensitive PAL5000-derived origin of replication (Guilhot et al., 1992). Six clones resistant to gentamicin and streptomycin were obtained after selection for the presence of the ΔmspA::aacC1 cassette and the absence of pMN226inv. The chromosomal DNAs of three clones were analysed by polymerase chain reaction (PCR). The primers MP14 and MP29 amplified a 1134 bp fragment from wild-type DNA and a 1778 bp fragment from the DNA of one recombinant clone (data not shown). This was the expected size increase of 644 bp after replacement of part of the mspA gene by the gentamicin resistance cassette. In addition, primers that are homologous to the deleted region, such as MP11 or ntmp01, gave only products from wild-type but not from recombinant DNA when used in combination with primers ctmp01 and MP04, respectively, which bind at the 3′ end of the mspA gene (data not shown). NruI cleaves the chromosomal DNA of wild-type M. smegmatis twice within the cloned mspA fragment, thereby releasing a 963 bp fragment (Fig. 1A). This fragment was present in wild-type chromosomal DNA but not in the DNA of the recombinant clone, as revealed by Southern hybridization using an mspA probe (Fig. 1B). Instead, an additional fragment of about 4400 bp was observed, in agreement with the deletion of one NruI site in the mutant. These data demonstrate that the mspA gene was exchanged by the ΔmspA::aacC1 cassette via homologous recombination. This ΔmspA mutant of M. smegmatis SMR5 was named M. smegmatis MN01. Although the ΔmspA mutant was viable, it is premature to conclude that mspA is not essential, because the expression of other porin genes could be activated in the mutant, as has been observed for quiescent porin genes in E. coli (Pugsley and Schnaitman, 1978; Prilipov et al., 1998).
MspA mRNA is not detectable in the mspA deletion mutant of M. smegmatis
Total RNA was prepared from wild-type M. smegmatis and M. smegmatis MN01 to examine whether the mspA gene is transcribed. By reverse transcription using random hexamers, total mRNA was converted into cDNA, which was amplified by the mspA-specific primers mspA-FP2 and mspA-RP2. No PCR products were obtained for the RNA of both strains without reverse transcription (data not shown). A DNA fragment with the correct length of 212 bp was observed for wild-type RNA after reverse transcription (RT), but not for RNA from the ΔmspA strain (Fig. 1C). In a control experiment, the same RNA was used after alkali treatment. No PCR product was detected in the RT–PCR assay using alkali-treated RNA in contrast to RNA supplemented with wild-type chromosomal DNA, which yielded a PCR product of the correct length (data not shown). This experiment indicated that DNA is not hydrolysed by alkali treatment and, consequently, that RNA was the template for DNA synthesis in the RT–PCR assay. Therefore, these data demonstrate that mspA mRNA is synthesized by wild-type M. smegmatis cells but not by the ΔmspA mutant.
Deletion of the mspA gene leads to the expression of another porin
Cells from wild-type M. smegmatis and the ΔmspA mutant were extracted with isotridecylpolyethyleneglycolether (Genapol) at high temperatures to compare the channel-forming activity of their cell walls. This extraction method yields about 85% pure MspA from wild-type M. smegmatis without further purification steps (Heinz and Niederweis, 2000). No channels were detected after addition of the extraction buffer alone in lipid bilayer experiments, whereas the membrane conductance increased by several orders of magnitude after the addition of both cell extracts (Fig. 2A). This indicated the presence of porins in both extracts. More than half the porins solubilized from wild-type cells had a single channel conductance of 4.6 nS (Fig. 2B) in 1 M potassium chloride, in agreement with earlier results (Niederweis et al., 1999). The single channel conductances for the porins solubilized from cells of the ΔmspA mutant were identical to that of MspA. However, the frequency of reconstitution events was roughly 50-fold higher for the wild-type extract compared with that of the ΔmspA mutant, as estimated from 19 single-channel experiments (Fig. 2A).
Gel electrophoresis revealed that high-temperature extracts of cells of the ΔmspA mutant with Genapol contained a major band with an apparent molecular mass of 100 kDa similar to MspA (Fig. 3A, lane 1). This protein was isolated from a preparative gel as described previously (Niederweis et al., 1999) and reconstituted as channels with a main conductance of 4.6 nS in lipid membranes (data not shown). This new porin of M. smegmatis was purified from extracts with the detergent n-octylpolyethylene oxide by anion exchange and gel filtration chromatography (Fig. 3A, lane 2). The yield was 50 µg porin l−1 culture, which is about 4.5-fold lower than that of MspA (Heinz and Niederweis, 2000). It had a high channel-forming activity in lipid bilayer experiments with a main conductance of 4.6 nS in potassium chloride, had the same electrophoretic mobility in denaturing polyacrylamide gels as purified MspA and was recognized by polyclonal antibodies to purified MspA from M. smegmatis in an immunoblot experiment (Fig. 3B). The new porin was resistant to organic solvents or boiling in the presence of 1% SDS, but dissociated into 20 kDa monomers after heating in 80% DMSO to 100°C as shown for MspA (Niederweis et al., 1999) (Fig. 3A). Sequencing by Edman degradation of the protein and of some peptides obtained after cleavage by the protease AspN revealed 20 N-terminal (GLDNELSLVDGQDRTLTVQQ) and six other amino acids (EPWNM) that were identical to the N- and the C-terminus of MspA. These results demonstrate that M. smegmatis MN01 expresses a porin with properties very similar to those of MspA.
Identification of other porin genes of M. smegmatis
Degenerated oligonucleotides ntmp02 and ctmp01 derived from the N- and C-termini of the new porin amplified a fragment of 550 bp from chromosomal DNA of M. smegmatis MN01. The PCR product was labelled with digoxigenin (DIG) and used to screen E. coli colonies harbouring a genomic library of M. smegmatis MN01. The plasmids pCS12 and pCS4 were isolated from two positive clones. They contained inserts of ≈ 19 kbp and 12 kbp of chromosomal M. smegmatis DNA. Sequencing revealed open reading frames (ORFs) of 648 bp and 621 bp in pCS12 and pCS4 with a homology of 91% and 75% to the mspA gene respectively. We named these genes mspC and mspD (M.smegmatisporin C and D). A blast search of the unfinished genome of M. smegmatis at The Institute for Genomic Research (http://www.tigr.org) using the MspA sequence yielded MspC, MspD and another protein that we called MspB. The N-terminal parts of these proteins showed all the characteristics of signal peptides (Pugsley, 1993). The mature proteins are remarkably similar with only two, four and 18 out of 184 amino acids being different in MspB, MspC and MspD compared with MspA (Fig. 4).
The high similarity of MspA and the porin of the ΔmspA mutant explains why both their channel conductance and their biochemical properties are indistinguishable. MALDI mass spectroscopy of the purified porin revealed a mass of 19 406 Da (Fig. 3C), which does not allow us to distinguish between MspB and MspC with predicted molecular masses of 19 370 and 19 371 Da respectively. As both the mspB and the mspC gene encoded all the amino acids determined by N-terminal sequencing of the purified porin of ΔmspA mutant, it is not clear which of these genes is expressed. No amino acids specific for MspD were observed. This is in agreement with the apparent absence of a ribosome binding site in front of the mspD gene, which might indicate that MspD is not synthesized in M. smegmatis.
The sequences of the porins of M. smegmatis did not show any significant match to any other sequence in the non-redundant GenBank database (574 979 sequences, 10/00, National Center for Biotechnology Information) using the basic local alignment search tool (blastp 2.1; Altschul et al., 1997) and default parameters. The similarity of MspB, C and D to MspA is > 90% and by far exceeds that of OmpF and OmpC, the two main general porins of E. coli, which share 61% identical amino acids (Mizuno et al., 1983). Thus, MspA, MspB, MspC and MspD are four members of a very homogeneous family of channel-forming proteins with distinct properties. At least two of these porins are extremely stable oligomers of 20 kDa subunits and presumably form the longest channels known to date because they have to span the 10-nm-thick mycolic acid layer of M. smegmatis (Brennan and Nikaido, 1995). The upstream and downstream sequences of the newly identified porin genes are different from those of the mspA gene, confirming their different location in the chromosome. These findings are in agreement with Southern blot analysis of NruI-digested chromosomal DNA, which revealed that M. smegmatis possesses four genomic fragments homologous to the mspA gene (Fig. 1B), extending our previous experiment, which detected three fragments with BamHI-digested chromosomal DNA using an mspA probe (Niederweis et al., 1999). However, it cannot be excluded that other porins less similar to the Msp porin family exist in M. smegmatis. In addition, it is not clear whether the Msp porins have different functions or whether some of them are back-up porins, such as NmpC and K in E. coli (Whitfield et al., 1983; Hindahl et al., 1984). In this regard, it should be noted that the exchange of one charged amino acid can drastically change the ion selectivity and thereby the physiological function of porins (Bauer et al., 1989). This might also be the case for the porins of M. smegmatis, as seven out of 18 amino acid exchanges involve charged amino acids. In addition, small differences in the single channel conductance in the range of 0.1–0.2 nS, as observed for OmpC and NmpC (Benz et al., 1985), were in the error range of our measurements and might therefore have escaped our notice. Furthermore, we do not know whether porin genes in addition to mspA are expressed in M. smegmatis. The fact that we did not detect any heterogeneity in our MspA preparations (Heinz and Niederweis, 2000) is probably not very indicative considering the high similarity of MspA, MspB and MspC.
The porin MspA is localized in the cell wall
The localization of MspA in the cell wall of M. smegmatis has not been demonstrated yet. In immunogold labelling experiments, we used the anti-MspA serum pAK MspA#813 (Niederweis et al., 1999), which specifically detects porins in total protein preparations of M. smegmatis as revealed in Western blots (data not shown). Ultrathin sections of wild-type cells of M. smegmatis embedded in acrylic resin and treated with these antibodies showed labelling at and close to the cell wall (Fig. 5A and B). Statistical analysis of five electron micrographs each containing two to 10 cell sections revealed between 150 and 400 gold labels µm−1 in the cell wall region using the anti-MspA antiserum (Fig. 5C). Up to 40 gold labels µm−1 cell wall region were detected using the preimmune serum. A similar density of about 50 gold labels µm−1 cytoplasmic area was observed with both anti-MspA antiserum and the preimmune serum. Thus, the labelling of the cell wall region of M. smegmatis with the anti-MspA antiserum was three- to eightfold above background and demonstrates that MspA is localized in the cell wall.
In an enzyme-linked immunosorbent assay (ELISA) (Burkovski, 1997), immobilized M. smegmatis cells did not give any signal using the preimmune serum, whereas antiserum pAK MspA#813 bound to both wild-type M. smegmatis and ΔmspA mutant cells (Fig. 6). This result shows that parts of MspA and the new porin are accessible at the cell surface of M. smegmatis. Furthermore, cell wall preparations of M. smegmatis using sucrose density gradients are highly enriched in MspA (data not shown). Thus, the results of three experiments show that MspA is localized in the cell wall of M. smegmatis, which is in agreement with its presumed function as a cell wall channel.
The mspA deletion mutant has a reduced cell wall permeability to a hydrophilic cephalosporine
The Zimmermann–Rosselet assay is one of the very few, theoretically correct methods for determining rates of diffusion through bacterial cell walls (Nikaido, 1985) and was therefore chosen to quantify the permeability of the cell wall of M. smegmatis and the ΔmspA mutant. In this assay, the permeability of the cell wall to the zwitterionic β-lactam antibiotic cephaloridine is measured using intact cells and periplasmic β-lactamases as a sink (Zimmermann and Rosselet, 1977; Jarlier and Nikaido, 1990). Cephaloridine (0.8 mM) was hydrolysed at a rate of 25.4 ± 4 nmol min−1 mg−1 by sonicated cells of M. smegmatis SMR5, which gives the total β-lactamase activity. In intact cells, cephaloridine hydrolysis was about fivefold slower with 4.4 ± 0.3 nmol min−1 mg−1, indicating that permeation through the outer membrane is rate limiting (Fig. 7A). Thus, the rate of cephaloridine hydrolysis is a direct measure of the permeability of the cell wall. The β-lactamase activity of the supernatant of M. smegmatis SMR5 cells was 0.3 ± 0.09 nmol min−1 mg−1, which is about 1% of the total β-lactamase activity and is in the error range of the rate of cephaloridine hydrolysis by intact cells. Therefore, the permeability coefficient of the cell wall of wild-type M. smegmatis for cephaloridine was calculated to (7.2 ± 1.4) × 10−7 cm s−1 in good agreement with the previously published value of (10 ± 1.2) × 10−7 cm s−1 (Trias and Benz, 1994). The total activity of the β-lactamases of the ΔmspA mutant M. smegmatis MN01 was identical to that of the parent strain (Fig. 7A). However, cephaloridine hydrolysis by intact cells of M. smegmatis MN01 was slower compared with the parent strain, yielding a permeability coefficient of (8.4 ± 2.2) × 10−8 cm s−1. Thus, deletion of the mspA gene reduced the permeability of the cell wall of M. smegmatis MN01 towards cephaloridine by a factor of about 9. This indicates an important role for MspA in the diffusion of hydrophilic solutes into M. smegmatis cells.
To verify whether this permeability defect was really caused by deletion of the mspA gene, we constructed a fusion of the mspA gene with the constitutive promoter psmyc of M. smegmatis. Transformation of M. smegmatis MN01 with the vector pMN014 containing the psmyc–mspA fusion did not change the total β-lactamase activity of the strain, but increased the permeability for cephaloridine to the wild-type level (Fig. 7A). Lipid bilayer measurements, protein analysis by gel electrophoresis and Western blot of detergent extracts of the complemented ΔmspA mutant showed that both the channel-forming activity of the extract and the intensity of the MspA band were increased to at least wild-type levels (data not shown). These results demonstrate that the permeability defects in M. smegmatis MN01, its low channel-forming activity and the lower number of porins in its cell wall are caused by deletion of the mspA gene. It is concluded that the MspA porin accounts for about 90% of the permeability of M. smegmatis cells towards cephaloridine.
Uptake of glucose by M. smegmatis is impaired if the mspA gene is deleted
To analyse whether mspA is important for the uptake of nutrients by M. smegmatis, we examined the accumulation of 14C-labelled glucose by intact cells. Uptake kinetics with 3.3 mM glucose showed a rapid saturation after about 5 min (Fig. 7B). The rate of glucose uptake by the ΔmspA mutant was 2.4-fold reduced compared with that of wild-type M. smegmatis. Complementation of the ΔmspA mutant with the mspA expression vector pMN014 increased uptake rates for glucose to wild-type levels (data not shown), confirming that deletion of mspA caused the permeability defect in M. smegmatis MN01. A series of uptake experiments with glucose concentrations ranging from 2 mM to 50 µM was performed to determine the apparent vmax and Km values of both strains. The effect of mspA deletion on glucose uptake was more pronounced at lower concentrations, in agreement with findings for Gram-negative bacteria (Ferenci, 1996), e.g. at 50 µM glucose, uptake was about 6.7-fold slower in the ΔmspA mutant compared with the wild type (data not shown). Data analysis using the Michaelis–Menten equation yielded vmax and Km values of 11.2 nmol mg−1 min−1 and 2.6 nM, and 1.3 nmol mg−1 min−1 and 1.1 nM for M. smegmatis wild type and the ΔmspA mutant respectively. Assuming that glucose penetrates into the cell by first diffusing passively through the cell wall and then being actively transported through the cytoplasmic membrane (Jarlier and Nikaido, 1990), one can calculate minimal values for the permeability coefficient P yielding 2.6 × 10−7 cm s−1 and 7.2 × 10−8 cm s−1 for wild-type M. smegmatis and the ΔmspA mutant respectively. Thus, deletion of mspA reduced the permeability of M. smegmatis for glucose fourfold to a value that is 200 000-fold lower than that determined for E. coli (Bavoil et al., 1977) and the lowest value reported so far for bacteria at all. These data demonstrate that diffusion through MspA is rate limiting for glucose uptake by M. smegmatis. The permeability coefficient of wild-type M. smegmatis for glucose is significantly lower than 3.7 × 10−6 cm s−1 as calculated from published data using eqn 1 (see Experimental procedures;Bai et al., 1978). It is obvious that, in our study, glucose uptake by M. smegmatis was slower, increased linearly with time only up to about 2 min and reached the saturation level much earlier. The reason for these different results is not known, but might be caused by the different growth conditions in both studies. In the earlier study, an unspecified strain of M. smegmatis was grown as surface culture (Bai et al., 1978) in contrast to liquid culture containing detergents used in this study. Our results show that glucose uptake in M. smegmatis is as slow as that measured for Mycobacterium chelonae (Jarlier and Nikaido, 1990). However, the permeability of M. chelonae for cephaloridine was sevenfold lower than that of M. smegmatis (Jarlier and Nikaido, 1990). This apparent contradiction might be explained by the different sizes of cephaloridine (415 g mol−1) and glucose (180 g mol−1), because it has been shown for E. coli that porin-mediated permeability differences are more pronounced for larger molecules (Nikaido and Rosenberg, 1983).
As the diffusion of such molecules as different as glucose and cephaloridine across the cell wall is impaired by deletion of the mspA gene, it is concluded that MspA provides a general diffusion pathway for hydrophilic molecules in M. smegmatis. In addition, uptake of fructose by cells from the ΔmspA mutant was significantly slower than that by wild-type cells (data not shown), supporting this assumption. The results of both transport assays correlated well with the reduced porin activity of detergent extracts and the lower porin content of the ΔmspA mutant. Thus, the physiological role of MspA in M. smegmatis is similar to that described for the porins OmpF and OmpC in E. coli many years ago (Nikaido and Nakae, 1979). It remains to be determined whether functional differences in the porins of the Msp family contribute to the permeability defect of the ΔmspA mutant or whether it results solely from a lower number of porins. The result that the ΔmspA mutant has lost about 90% and 75% of its permeability for cephaloridine and glucose, respectively, indicates that MspA is the main porin of M. smegmatis and was only partially replaced by fewer porin molecules in the cell wall of the ΔmspA mutant.
It has been assumed that the low cell wall permeability of mycobacteria for nutrients might contribute to their slow growth (Ratledge, 1982). This hypothesis was supported by theoretical considerations for M. chelonae (Jarlier and Nikaido, 1990). To analyse whether deletion of mspA affected the growth rate, wild-type M. smegmatis and the ΔmspA mutant were grown in minimal medium supplemented with glucose as the sole carbon source in concentrations ranging from 10 µM to 10 mM. However, growth of both strains in the presence of glucose up to 100 µM was very slow, stopped at an OD600 of about 0.1 and did not differ significantly. At higher glucose concentrations, growth of both wild-type M. smegmatis and the ΔmspA mutant was supported up to an OD600 of about 0.3 (Fig. 8). The growth rates of both strains were similar despite the slower glucose uptake of the ΔmspA mutant. In the presence of 1 mM glucosamine (Fig. 8) or 1 mM succinate (data not shown), a small, but significant retardation of growth was observed for the ΔmspA mutant compared with wild-type M. smegmatis. These results show that even the very low outer membrane permeability of the ΔmspA mutant allows a glucose influx sufficiently rapid for an unaffected growth rate and therefore argue against the low cell wall permeability as a determinant of the slow growth of wild-type M. smegmatis. This result is not surprising considering the presence of at least three additional porins in M. smegmatis and the fact that a severe growth defect in E. coli was only observed in mutants that lost both main porins, OmpF and OmpC (Bavoil et al., 1977). In addition, a sugar-specific porin might exist in M. smegmatis, such as LamB in E. coli, which was shown to be important for the growth of E. coli under limiting sugar concentrations even in the presence of OmpF and OmpC (Death and Ferenci, 1993). However, the slightly reduced growth rate of the ΔmspA mutant with glucosamine might indicate that deletion of mspA reduced the outer membrane permeability of M. smegmatis to a level that is just growth rate limiting. Further experiments are needed to check whether this effect is caused by slower transport of glucosamine across the mycolic acid layer or different uptake rates by cytoplasmic transporters compared with glucose.
Recently, it has been shown that synthesis of mycolic acids is essential for the mycobacterial cell wall to function as an effective permeability barrier to hydrophobic molecules (Jackson et al., 1999; Liu and Nikaido, 1999; Wang et al., 2000). This study demonstrates that MspA is the main hydrophilic pathway in the cell wall of M. smegmatis and therefore provides the first experimental evidence that mycobacterial porins are the major determinants of permeability to hydrophilic molecules. This result is the first step in understanding the exceptionally low permeability of mycobacteria to hydrophilic molecules at a molecular level.
Bacterial strains and growth conditions
Mycobacterium smegmatis SMR5 is a streptomycin-resistant mutant of M. smegmatis mc2155 (kindly provided by Peter Sander; Sander et al., 1995) and was grown in Middlebrook 7H9 medium (Difco Laboratories) at 37°C supplemented with 0.2% glycerol, 0.05% Tween 80. Escherichia coli DH5α was used for all cloning experiments and was routinely grown in LB medium at 37°C unless otherwise stated. Antibiotics were used when required at the following concentrations: ampicillin (100 µg ml−1 for E. coli), kanamycin (30 µg ml−1 for E. coli; 10 µg ml−1 for M. smegmatis), hygromycin (200 µg ml−1 for E. coli; 50 µg ml−1 for M. smegmatis), gentamicin (15 µg ml−1) and streptomycin (400 µg ml−1).
Construction of plasmids
The plasmid pMycVec1 (to be published elsewhere) was digested with FseI, treated with Klenow polymerase to remove protruding 3′ ends, then cut with KpnI. This vector fragment of pMycVec1 was ligated with the rpsL gene, which was cut out of ptrpA-1-rpsL+ (kindly provided by Peter Sander; Sander et al., 1995) with SmaI and KpnI, to give pMN210. The pAL5000 origin of replication of pMN210 was replaced by its temperature-sensitive (ts) variant from pCG63 (kindly provided by Brigitte Gicquel; Guilhot et al., 1992). Therefore, pMN210 was cut with ApaI, and the ts pAL5000 origin was excised from pCG63 with EcoRV–KpnI. The ends of both DNA fragments were converted to blunt ends with Klenow polymerase and ligated. The resulting plasmid, pALEX1, contained two counterselectable markers (the rpsL gene and a ts pAL5000 origin of replication) and was used as a vector for recombination in mycobacteria. In the first step to construct a deletion of the mspA gene, a 2.8 kbp genomic fragment from M. smegmatis containing the mspA gene was excised from pPOR6 by SpeI and EcoRV, converted to blunt ends with Klenow polymerase and cloned into the filled-in SalI site of pBluescript-II KS+ (Stratagene) to give pBlue-mspA. A total of 536 bp containing the putative promoter, Shine–Dalgarno (SD) sequence, translation start and part of the mspA gene encoding the signal peptide and 54 amino acids of the mature protein was deleted from the genomic fragment of pBlue-mspA by cutting with FseI and StyI. The vector fragment was converted to blunt ends with T4 DNA polymerase and ligated with a 1180 bp gentamicin resistance cassette that was amplified from pMV361 (Stover et al., 1991) using the primers GenR-fwd and GenR-rev to give pMN224. The genomic fragment that contained the ΔmspA::aacC1 cassette was cut out of pMN224 with ApaI and SacI, treated with T4 DNA polymerase to get blunt ends and cloned into the single PmeI site of pALEX1 to give the mspA knock-out vector pMN226inv.
In order to construct a mycobacterial expression vector for the mspA gene, the mspA gene was amplified from pPOR6 with the oligonucleotides MP-fwd, which introduced a PacI site, and MP-rev, which has a half-site of the SwaI site (Table 1). The PCR fragment was digested with PacI, purified by preparative gel electrophoresis and ligated with the PacI- and SwaI-digested pMS2 DNA to give pMN006. A strong promoter from M. smegmatis was isolated from pUV15 (kindly provided by Sabine Ehrt) as a PmeI–PacI fragment and cloned via the same restriction sites into pMN006 to yield the mspA expression vector pMN014 (details to be published elsewhere). All plasmid constructions were verified by restriction enzyme digestion and double-stranded DNA sequencing.
Table 1. Oligonucleotides used in this study.
Sequence (5′ to 3′)
. Positions marked in bold contained mixed nucleotides. ntmp01: 30% T at position 3, 25% C at position 6; ctmp01: 50% C at position 13.
Construction of an mspA deletion mutant of M. smegmatis
M. smegmatis SMR5 was transformed with pMN226inv, which contains the ΔmspA::aacC1 cassette. Gentamicin is the marker for positive selection of the ΔmspA::aacC1 cassette, whereas growth of M. smegmatis that still contains the plasmid pMN226inv should be inhibited in the presence of streptomycin and at 39°C, which is the non-permissive temperature for plasmid replication. After selection on 7H10 plates containing gentamicin and streptomycin at 39°C for 4 days, six clones were obtained. The colonies of these clones were picked and streaked onto 7H10 plates containing gentamicin and streptomycin with and without kanamycin to confirm their GenR StrR KanS phenotype and to check for loss of pMN226inv. This procedure was repeated twice with each strain. Three of these clones were grown in 7H9 medium containing gentamicin and streptomycin to an OD600 of 1, and chromosomal DNA was prepared for Southern blot and PCR analysis as described. One strain contained the chromosomal deletion of the mspA gene and was subsequently called M. smegmatis MN01.
Southern blot and PCR analysis of chromosomal DNA
Chromosomal DNA was isolated from M. smegmatis SMR5 and M. smegmatis MN01 as described previously (van Soolingen et al., 1991). For Southern blot analysis, 15 µg of chromosomal DNA was digested with BamHI and NruI overnight, separated on a 1% agarose gel and transferred in 20× SSC (3 M NaCl, 0.3 M sodium citrate, pH 7.0) to an uncharged nylon membrane (Gibco BRL). The DNA was cross-linked to the membrane by UV irradiation and prehybridized with salmon sperm DNA using standard protocols (Ausubel et al., 1987). The mspA probe was synthesized by PCR from pPOR6 using the primers MP01 and MP29. This 531 bp DNA fragment is homologous to 441 bp of the 3′ end of the mspA gene and was labelled by incorporation of biotin-7-dATP in a nick translation reaction (BioNick kit; Gibco BRL). The membrane was hybridized with the probe overnight at 43°C. Then, it was washed twice for 5 min at room temperature with 2× SSC, 0.1% SDS and 0.2× SSC, 0.1% SDS and twice for 5 min at 55°C with 0.16× SSC, 0.1% SDS. The membrane was blocked with 3% BSA in TBST for 60 min at 65°C. The hybridized biotin-labelled probe was detected using a photoreaction of 4-methoxy-4-(3-phosphatphenyl)-spiro-[1.2-dioxetan-3′,2′-adamantan) (PPD) after dephosphorylation by a conjugate of streptavidine and alkaline phosphatase (PhotoGENE kit; Gibco BRL). After 1 h at 37°C in the dark, the membrane was exposed to X-AR-5 films (Kodak).
For each PCR, 250 ng of chromosomal DNA from M. smegmatis SMR5 and MN01 was used as template, which was amplified with one of the following three primer pairs: MP14/MP29, MP11/ctmp01 and ntmp01/MP04. The PCR consisted of 30 cycles with the following steps: 30 s denaturation at 96°C, 30 s annealing and 1 min extension at 72°C. Specific amplification of the fragments with the expected sizes was obtained at the following annealing temperatures: 62°C for MP14/MP29 and MP11/ctmp01 and 52°C for ntmp01/MP04. One-tenth of the reaction volume was loaded on a 1% agarose gel.
RNA preparation and RT–PCR
M. smegmatis was grown in 5 ml cultures to an OD600 of 0.5. Cells were harvested by centrifugation and lysed by agitation with glass beads (FastRNA Tubes-Blue) in a FastPrep FP120 bead beater apparatus (Bio-101). RNA was isolated from cell lysates using the RNeasy kit according to the manufacturer's protocol for the isolation of total RNA from bacteria (Qiagen). Purified RNA was eluted with 50 µl of RNase-free water and treated with 10 U of RNase-free DNase I (Boehringer Mannheim) in T4 DNA ligase buffer (Gibco BRL) for 1 h at 37°C. The integrity of the RNA preparation was verified by the presence of two sharp rRNA bands by agarose gel electrophoresis. The reverse transcription (RT) reaction was performed using the reverse transcription reagents master mix (Perkin-Elmer). Total RNA (2 µg) was transcribed by Multiscribe reverse transcriptase into cDNA with random hexamers for 45 min at 48°C. One-twentieth of the volume of the RT reaction containing 100 ng of RNA was used after the addition of the mspA-specific primers mspA-FP2 and mspA-RP2 for PCR. Primer annealing was performed at 59°C for 30 s.
RNA hydrolysis was carried out as described previously (Donis-Keller et al., 1977). One microlitre of 1 M NaOH was added to 4 µg of total RNA in a total volume of 15 µl and incubated for 1 h at 72°C before neutralization with 1 µl of 1 M HCl. After ethanol precipitation, part of the sample corresponding to 100 ng of RNA was used in the subsequent PCR. Precipitation of nucleic acids by ethanol and RNA hydrolysis was checked by adding 100 ng of chromosomal DNA from M. smegmatis SMR5 to the hydrolysis reaction.
Preparation of detergent extracts from M. smegmatis and lipid bilayer experiments
Porins were selectively extracted from whole cells of M. smegmatis as described previously (Heinz and Niederweis, 2000). For the analysis of whole-cell extracts, 10 mg of M. smegmatis SMR5 or MN01 cells (wet weight) was washed with phosphate-buffered saline (PBS), resuspended in 150 µl of PG05 buffer and incubated for 30 min at 100°C. The samples were cooled on ice for 10 min, centrifuged at 4°C for 10 min and diluted 100-fold in PG05 buffer for lipid bilayer measurements, which were performed as described previously (Niederweis et al., 1999). The volume of the supernatant was reduced from 150 µl to 10 µl by evaporation for protein gel electrophoresis.
Purification and biochemical analysis of the porin of the ΔmspA mutant
For large-scale preparations of the porin of the ΔmspA mutant, selective extraction of 15 g of M. smegmatis MN01 with POP05 buffer and protein purification was performed as described previously (Heinz and Niederweis, 2000). Porin preparations were analysed for channel-forming activity using the lipid bilayer method and by SDS–PAGE. Protein concentrations were determined using bicinchoninic acid (BCA) (Smith et al., 1985) and bovine serum albumin (BSA) as standard protein. Protein gel electrophoresis, N-terminal amino acid sequencing and mass spectrometry were carried out as described for MspA (Niederweis et al., 1999; Heinz and Niederweis, 2000).
Cloning and sequencing of the mspC and mspD genes
Chromosomal DNA of M. smegmatis MN01 was digested with PstI and cloned into the PstI site of pMS2. The genomic library was amplified in E. coli DH5α. Colony hybridization experiments were performed according to the manufacturer's protocol (Boehringer Mannheim). The probe was synthesized by PCR from chromosomal DNA of M. smegmatis MN01 using the primers ntmp01 and ctmp02 (see above). Sequencing revealed that these primers specifically amplified the mspC gene. The PCR fragment was denatured and labelled with digoxygenin (DIG) using Klenow polymerase and random hexamers as described by the manufacturer (Boehringer Mannheim). Hybridization was carried out in the presence of 450 ng of DIG-labelled PCR fragment at 42°C overnight. The membrane was washed twice for 5 min at room temperature with 2× SSC, 0.1% SDS and twice for 15 min at 68°C with 0.1× SSC, 0.1% SDS. The probe was detected using the DIG nucleic acid detection kit (Boehringer Mannheim). The plasmids pCS4 and pCS12 were isolated from one clone, which hybridized with the probe and contained inserts of chromosomal M. smegmatis DNA of ≈ 12 kbp and 19 kbp. Part of these inserts was sequenced using the Dye Terminator cycle sequencing kit (Perkin-Elmer) and an Applied Biosystems 310 sequencer (Perkin-Elmer) by primer walking. Both strands were sequenced at least twice. The accession numbers for the EMBL Nucleotide Sequence Database are AJ299735 for mspC and AJ300774 for mspD.
Proteins from bacterial extracts, lysates or purified proteins were transferred to a nitrocellulose or polyvinylidene difluoride (PVDF) membrane using a standard protocol (Ausubel et al., 1987) and detected by the rabbit antiserum pAK MspA#813, which appears to be specific for mycobacterial porins (Niederweis et al., 1999). Horseradish peroxidase coupled to an anti-rabbit antibody oxidized luminol (ECL plus kit; Amersham) whose chemoluminescence was detected by X-AR-5 films (Kodak).
M. smegmatis was prepared for immunogold electron microscopy by the method of progressive lowering of temperature (PLT) essentially as described previously (Glauert and Lewis, 1998). Cell suspensions were centrifuged at 5000 r.p.m. for 5 min using a Sorvall SS35 rotor to remove aggregated cell material. The supernatant was mixed with the same volume of a solution containing 8% paraformaldehyde and 1% glutaraldehyde dissolved in 100 mM phosphate buffer, pH 6.5, 0.1 mM EDTA, 150 mM NaCl plus 0.05% Tween 80 and fixed for 2 h. Cells were sedimented at 10 000 r.p.m. for 5 min in a Beckman microfuge 11, and the pellets were resuspended in low gelling agarose at 60°C and cut into cubes of about 1 mm3 in size after polymerization at room temperature. Samples were dehydrated through a graded ethanol series from 30% at −10°C, via 50% (−20°C) and 70% (−25°C) to 100% at −35°C for 5–24 h. Absolute ethanol was diluted with buffer in the absence of any aldehyde. The samples were embedded in a polar acrylic resin (LR white; London Resin Company) and dissolved in ethanol using concentrations of LR white of 30% at −30°C, 50% (−20°C), 70% (−20°C) and 100% at −15°C. Incubations were between 3 h and 17 h. Finally, polymerization was performed at 55°C for 3 days. Sections were cut on a Reichert-Jung Ultracut E ultratome to silver/gold interference and picked up on carbon-coated Ni grids.
Ultrathin sections were blocked with PBG1 solution, consisting of 50 mM glycine in 12 mM phosphate, pH 7.4, plus 150 mM NaCl (PBS), twice for 5 min at room temperature followed by double treatment with 0.2% fish gelatine in PBS (PBG2) for 10 min. Before exposure to a polyclonal antibody diluted in blocking solution PBG1 at a ratio of 1:50 for 2 h at room temperature, the sections were rinsed with PBG1. The grids were washed five times for 3 min in PBG1 before incubation with the immunogold conjugate, a 1:100 dilution of 10 nm colloidal gold-labelled goat anti-rabbit antibody, for 1 h. The grids were washed a further three times for 5 min in blocking solution PBG1 and once for 5 min in distilled water before drying for 10 min and staining with an aqueous solution of 2% uranyl acetate for 10 min. Afterwards, the grids were washed with water and dried again before staining for 5 min with lead citrate containing 26.6 mg of lead citrate, 35.2 mg of sodium citrate and 0.16 mmol of NaOH ml−1. Finally, the grids were rinsed with 20 mM NaOH followed by water. After drying, the samples were backed with 0.5 nm carbon in a Balzers T-80 vacuum device. Controls of preimmune serum and conjugate, as well as conjugate only, were incorporated to ensure that low background labelling and high specificity of labelling were achieved. Sections were inspected in a Philips EM 420 electron microscope at an acceleration voltage of 80 kV, and images were recorded at a primary magnification of 35 000-fold. For statistical analysis the cell wall region was defined as the cell wall including 20 nm on each side.
Enzyme-linked immunosorbent assay (ELISA)
Cells were grown to an OD600 of about 0.6, harvested by centrifugation, washed twice in PBS (see above) containing 0.05% Tween 80 and resuspended in 50 mM NaHCO3, pH 9.6, to yield a cell concentration of about 1 × 108 cells ml−1. Aliquots (200 µl) of this cell suspension or dilutions thereof were transferred to wells of microtitre plates (NUNC-Immuno MaxiSorp Surface; Nalge Nunc International). After incubation overnight at 4°C, wells were washed twice with 200 µl of TBST buffer containing 50 mM Tris-HCl, pH 7.8, 150 mM NaCl, 1 mM MgCl2 and 0.05% Tween 80. The remaining protein binding sites were blocked with 200 µl of 3% powdered skim milk in TBS for 1.5 h at room temperature. The wells were washed three times with 200 µl of TBST. Antisera were diluted 1:3000. Incubation with preimmune serum or the first antibody (pAK MspA#813) was carried out for 1.5 h at room temperature. The wells were washed five times with 200 µl of TBST. Anti-rabbit antibody alkaline phosphatase conjugate (Sigma) was diluted 1:10 000 in TBS and allowed to bind the first antibody (Sigma) at room temperature for 1 h. After six washing steps with 200 µl of TBST, 200 µl of a solution of 1 mg of p-nitrophenyl phosphate (Merck) ml−1 0.1 M glycine, pH 10.4, 1 mM ZnCl2, 1 mM MgCl2 was added well−1 and incubated for 1.5 h at room temperature. The reaction was stopped by the addition of 50 µl of 3 M NaOH. Absorption at 405 nm was measured with a microplate reader (Emax precision microplates reader; Molecular Devices). Increasing the concentration above 108 cells ml−1 well−1 caused a decreased absorbance probably because of a greater loss of cells.
Quantification of the outer membrane permeability of M. smegmatis to cephaloridine
The permeation of cephaloridine through the mycolic acid layer of M. smegmatis was measured spectrophotometrically using the method of Zimmermann and Rosselet (1977) as modified by Jarlier and Nikaido (1990). M. smegmatis SMR5 and MN01 were grown to an OD600 of 1, harvested by centrifugation for 5 min at 3000 g at room temperature, washed with PBS and resuspended in 2.5 mM PIPES (pH 6.5) to a concentration of 80 mg of cells (wet weight) ml−1. This cell suspension (100 µl) was mixed with 400 µl of a 2.5 mM PIPES (pH 6.5) buffer containing 1 mM cephaloridine (Sigma). A sample of 300 µl of this mixture was quickly transferred to a cuvette with 1 mm light path, and the ODs at 260 nm and at 241 nm, the isosbestic wavelength of cephaloridine, were recorded for 40 min at 25°C. In order to determine the activity of the periplasmic β-lactamases of M. smegmatis, 160 mg of cells (wet weight) resuspended in 1 ml of 2.5 mM PIPES (pH 6.5) was broken by sonication, and the hydrolysis of cephaloridine by the supernatant corresponding to 80 mg (wet weight) of broken cells was measured as described above. The maximal rate for the β-lactamases of M. smegmatis of the sonicate was 25.4 ± 4 nmol min−1 mg−1 cells (dry weight). The mean dry weight of the cells was 0.57 mg and 0.78 mg for wild-type M. smegmatis and the ΔmspA mutant respectively. Special care was taken in order to prevent cell lysis by mechanical stress, which would release periplasmic β-lactamases into the supernatant. This was checked by measuring cephaloridine hydrolysis by the supernatant of 80 mg of intact cells (wet weight) as described above. Permeability coefficients P were calculated according to the method of Zimmermann and Rosselet (1977) using a KM of 146 µM for the β-lactamases of M. smegmatis (Trias and Benz, 1994) and an approximation of 132 cm2 mg−1 (dry weight) as the surface area/weight ratio for mycobacteria (Jarlier and Nikaido, 1990).
Glucose uptake measurements
M. smegmatis SMR5, MN01 and MN01 complemented with an episomal copy of mspA (pMN014) were grown to an OD600 of 1.5, harvested by centrifugation for 10 min at 3000 g at 4°C, washed twice with 2 mM PIPES (pH 6.5), 0.05 mM MgCl2 and resuspended in the same buffer. [14C]-glucose (specific activity 311 mCi mmol−1; Amersham) was mixed with glucose and added to the cell suspension to obtain final concentrations ranging from 50 µM to 3.3 mM. The mixture was incubated at 25°C and 37°C as specified, and 1 ml samples were removed at times ranging from 15 s to 16 min. The cells were filtered through a 0.45 µm pore size membrane filter (Sartorius), washed with 0.1 M LiCl and counted in a liquid scintillation counter. One millilitre of each cell suspension was dried and weighed to determine the concentration of the cells used in the assay. The mean dry weight was 0.84 ± 0.02 mg. The uptake of glucose was expressed as nmol mg−1 (dry weight) cells. The uptake rate was determined by fitting a straight line to at least the first three data points (from 15 to 45 s). The correlation coefficients for all fits were greater than 0.97, indicating that the data did not deviate significantly from the straight line. Excellent fits of the uptake rates determined at different glucose concentrations were obtained using the Michaelis–Menten equation, yielding KM and vmax values for the overall transport. This data analysis was confirmed using a Lineweaver–Burk plot for wild-type M. smegmatis. A minimal estimate of the permeability coefficient was obtained using the equation:
which was derived earlier (Jarlier and Nikaido, 1990) by assuming that the substrate diffuses passively through the cell wall and is actively transported through the cytoplasmic membrane at a higher rate. Thus, the substrate concentration in the periplasm can be neglected compared with the initial substrate concentration in the buffer. We used 132 cm2 mg−1 (dry weight) as an estimation of the surface area A to weight ratio for mycobacteria as published previously (Jarlier and Nikaido, 1990).
M. smegmatis SMR5 and MN01 were grown as 5 ml precultures for 36–40 h in MMT minimal medium (60 g l−1 Na2HPO4, 30 g l−1 KH2PO4, pH 7.4, 5 g l−1 NaCl, 10 g l−1 NH4Cl, 2 mM MgSO4, 0.1 mM CaCl2, 0.05% Tween 80) containing 10 mM glucose. The cells were washed twice in MMT and diluted in 100 ml of MMT supplemented with 0.01, 0.1, 1 and 10 mM glucose, 1 mM succinate and 1 mM glucosamine as carbon sources. Growth rates were determined in triplicate by OD600 measurements.
Polyacrylamide gels were dried using the DryEase minigel drying system (Novex) and scanned (UC840 Max Vision; Umax). The images were imported into photoshop 5.0 (Adobe), and brightness and the global gradation curve were adjusted to reduce the background. No parts of the gels were changed individually.
We thank Wolfgang Hillen for generous support, Peter Sander, Erik Boettger, Sabine Ehrt, Lee W. Riley and Brigitte Gicquel for provision of plasmids and strains, and Brigitte Kühlmorgen for skilful technical assistance with immunoelectron microscopy. Preliminary sequence data were obtained from The Institute for Genomic Research website at http://www.tigr.org. This work was supported by the Deutsche Forschungsgemeinschaft (NI 412).
Present addresses: †Institut für Biochemie, Universität Zürich, Winterthurerstr. 190, CH-8057 Zürich, Switzerland.
‡Institut für Biochemie, Friedrich-Alexander-Universität Erlangen-Nürnberg, Fahrstr. 17, D-91054 Erlangen, Germany.
§Max-Planck-Institut für Biochemie, Abteilung Molekulare Strukturbiologie, Am Klopferspitz 18a, D-82152 Martinsried, Germany.