Deletion of the genes encoding the MtrA–MtrB two-component system of Corynebacterium glutamicum has a strong influence on cell morphology, antibiotics susceptibility and expression of genes involved in osmoprotection
Institut für Biochemie der Universität zu Köln, D-50674 Köln, Germany.
The MtrAB two-component signal transduction system is highly conserved in sequence and genomic organization in Mycobacterium and Corynebacterium species, but its function is completely unknown. Here, the role of MtrAB was studied with C. glutamicum as model organism. In contrast to M. tuberculosis, it was possible to delete the mtrAB genes in C. glutamicum. The mutant cells showed a radically different cell morphology and were more sensitive to penicillin, vancomycin and lysozyme but more resistant to ethambutol. In order to identify the molecular basis for this pleiotropic phenotype, the mRNA profiles of mutant and wild type were compared with DNA microarrays. Three genes showed a more than threefold increased RNA level in the mutant, i.e. mepA (NCgl2411) encoding a putative secreted metalloprotease, ppmA (NCgl2737 ) encoding a putative membrane-bound protease modulator, and lpqB encoding a putative lipoprotein of unknown function. Expression of plasmid-encoded mepA in Escherichia coli led to elongated cells that were hypersensitive to an osmotic downshift, supporting the idea that peptidoglycan is the target of MepA. The mRNA level of two genes was more than fivefold decreased in the mutant, i.e. betP and proP which encode transporters for the uptake of betaine and proline respectively. The microarray results were confirmed by primer extension and RNA dot blot experiments. In the latter, the transcript level of genes involved in osmoprotection was tested before and after an osmotic upshift. The mRNA level of betP, proP and lcoP was strongly reduced or undetectable in the mutant, whereas that of mscL (mechanosensitive channel) was increased. The changes in cell morphology, antibiotics susceptibility and the mRNA levels of betP, proP, lcoP, mscL and mepA could be reversed by expression of plasmid-encoded copies of mtrAB in the ΔmtrAB mutant, confirming that these changes occurred as a consequence of the mtrAB deletion.
Two-component signal transduction systems are widespread in bacteria and mediate adaptations to changing environmental conditions (Hoch and Silhavy, 1995). The sensor kinase usually responds to a certain stimulus by autophosphorylation of a histidine residue and the phosphoryl group is subsequently transferred to an aspartate residue of the cognate response regulator, which then mediates changes in gene expression or cell behaviour. The two-component system composed of the sensor kinase MtrB and the response regulator MtrA was first described in the human pathogen Mycobacterium tuberculosis. The mtrA–mtrB genes of this species were identified and cloned by heterologous hybridization with a Pseudomonas aeruginosa phoB probe (Via et al., 1996). The purified MtrA protein could be phosphorylated using the heterologous sensor kinase CheA from Escherichia coli as phosphoryl donor, confirming that it represents a response regulator (Via et al., 1996). Attempts to disrupt the mtrA gene in M. tuberculosis H37Rv failed, except when a plasmid-encoded intact mtrA copy was present. Thus, mtrA appears to be an essential gene for M. tuberculosis and therefore should be expressed under the growth conditions used during mutant construction. In fact, S1 nuclease mapping and primer extension analysis revealed the presence of mtrA mRNA and showed furthermore that the transcriptional start site of mtrA is identical with the translational start site (Zahrt and Deretic, 2000). This situation has also been observed in a variety of other genes from mycobacteria and corynebacteria (Moll et al., 2002). Using a plasmid-encoded mtrA-gfp (green fluorescent protein) transcriptional fusion, it was shown that mtrA was significantly induced in the vaccine strain M. bovis BCG after infection of a murine macrophage cell line, whereas mtrA expression in the virulent strain M. tuberculosis H37Rv was strong even before infection and no further induction was observed after infection of murine or human macrophages (Via et al., 1996; Zahrt and Deretic, 2000).
Because MtrA contains a helix–turn–helix motif in its carboxy-terminal domain, it presumably functions as transcriptional regulator. However, its target genes as well as the signals recognized by MtrB, the presumptive sensor kinase of MtrA, are completely unknown. In this work, the question on the function of the MtrAB two-component system was addressed using Corynebacterium glutamicum as model organism. This species is a Gram-positive soil bacterium which like the mycobacteria belongs to the suborder Corynebacterineae within the Actinomycetales (Stackebrandt et al., 1997). The members of this suborder are characterized by a unique cell envelope containing a core mycolyl-arabinogalactan peptidoglycan complex (Crick et al., 2001). The mycolic acid residues form an outer hydrophobic layer, resembling in function the outer membrane of Gram-negative bacteria (Bayan et al., 2003).
In the past, scientific research on C. glutamicum was primarily driven by its use for large-scale industrial amino acid production, in particular l-glutamate and l-lysine. Meanwhile, this species is also of increasing interest as a non-pathogenic model organism to study selected topics that are common to its pathogenic relatives C. diphtheriae, M. tuberculosis and M. leprae, e.g. cell envelope biosynthesis (Gibson et al., 2003). Many of the genes present in the completely sequenced 3.3 Mb genome of C. glutamicum (Ikeda and Nakagawa, 2003; Kalinowski et al. 2003) are highly conserved in sequence and gene order within the other members of the Corynebacterineae. This is also true for the mtrAB genes (Fig. 1). In contrast to M. tuberculosis, the mtrAB genes of C. glutamicum could be deleted and a comparison of the ΔmtrAB mutant with the wild type by phenotypic characterization and transcriptional profiling indicated that MtrAB is involved in cell wall metabolism and influences the expression of genes involved in osmoprotection.
Conservation of the mtrAB locus in the genomes of Corynebacterium and Mycobacterium species
In contrast to their conserved kinase domains, the sensor domains of histidine kinases are highly variable, reflecting the diverse stimuli recognized by these domains. When searching for two-component systems that perform similar functions, it is helpful to perform searches for sequence similarity only with the sensor domain. In the case of the histidine kinase MtrB from M. tuberculosis, the sensor domain contains two putative transmembrane helices extending from residues 42–62 and 214–234, which enclose a presumably periplasmic domain of 151 residues. When this domain was used for blastp searches (Altschul et al., 1990), proteins with significant similarity (E-values ≤ e−10) were found to be encoded in the genomes of M. bovis, M. marinum, M. leprae, M. avium ssp. paratuberculosis, C. diphtheriae, C. efficiens, C. glutamicum and Rhodococcus sp. strain I24. The genomic vicinity of some of the corresponding mtrAB loci is shown in Fig. 1. In all cases, downstream of mtrB a gene named lpqB is found, whose putative GTG start codon overlaps by two nucleotides with the mtrB stop codon in the case of C. glutamicum. Although not characterized functionally, in silico analysis revealed that all LpqB proteins are probably lipoproteins as they contain an amino-terminal signal peptide for Sec-dependent translocation followed by a lipobox (VAGC in C. glutamicum LpqB) for glyceryltransferase-dependent lipidation of the cysteine residue and subsequent processing by signal peptidase II (Sutcliffe and Harrington, 2002).
Construction and growth properties of a C. glutamicum ΔmtrAB mutant
In order to characterize the MtrAB two-component system, the non-pathogenic C. glutamicum was used as model organism. The MtrA and MtrB proteins of this species show 74% and 52% overall sequence identity to the corresponding proteins from M. tuberculosis. The C. glutamicum mtrA gene codes for a protein of 226 amino acyl residues consisting of an amino-terminal receiver domain and a carboxy-terminal DNA-binding domain belonging to the OmpR family. The aspartate residue which presumably represents the phosphorylation site is located at position 58. The first potential start codon (CTG) of mtrB is located 87 bp downstream of the mtrA stop codon and encodes a histidine kinase of 503 amino acyl residues. A second putative start codon (GTG) leads to a protein of 499 residues. The following numbers refer to MtrB consisting of 503 residues. MtrB is predicted to have two transmembrane helices, extending from residues 9–29 and 175–195 which enclose a periplasmic domain of 145 amino acids. The histidine residue most probably phosphorylated is found at position 266.
For functional analysis, a deletion mutant was constructed using a two-step homologous recombination method (Niebisch and Bott, 2001) which lacked the mtrAB coding regions except for the six 5′-terminal mtrA codons and the twelve 3′-terminal mtrB codons. The deletion was confirmed by polymerase chain reaction (PCR) and Southern blot analysis. For PCR, primers Delta-mtrAB-out-fw and Delta-mtrAB-out-rv were used which anneal outside the deleted region. As expected, a 3051 bp fragment was amplified from chromosomal DNA of the wild type, whereas in the ΔmtrAB mutant an 849 bp fragment was obtained (data not shown). Southern blot analysis with a 2968 bp digoxigenin-labelled probe (PCR product obtained with the oligonucleotides Delta-mtrAB-1 and Delta-mtrAB-4 and genomic C. glutamicum DNA as template) revealed two fragments of 2.9 and 7.6 kb in a DraI/HindIII digest of chromosomal DNA of the wild type and, as predicted, a single fragment of 8.2 kb in the ΔmtrAB mutant (data not shown).
Growth of the C. glutamicumΔmtrAB strain was compared with that of the wild-type strain in minimal media. A typical growth experiment in CGXII minimal medium with 4% glucose as sole carbon source is shown in Fig. 2. Although both strains exhibited identical growth rates during the early exponential phase, the mtrAB deletion mutant entered the stationary phase at a much lower cell optical density (OD600 24) than the wild type (OD600 38). The result of the OD measurements was confirmed independently by the determination of the cell dry weight (data not shown). In order to elucidate the reason for the observed growth differences, the pH and the glucose concentration of the stationary phase cultures were measured. While the pH of the wild-type medium had changed from pH 7.0 to pH 6.5 during cultivation (Fig. 2A), that of the ΔmtrAB culture had dropped to pH 4.5 (Fig. 2B), even though CGXII medium has a high buffering capacity (200 mM MOPS, 83 mM urea, 13 mM phosphate). In the wild-type culture, glucose had been completely consumed, whereas the ΔmtrAB culture still contained 70 mM residual glucose (initially 212 mM), showing that the reduced growth of this strain does not result from carbon limitation. When the pH of the ΔmtrAB culture was readjusted to pH 7 during growth, the culture reached the same density as the wild type, showing that the growth arrest of the ΔmtrAB mutant at a low cell density is caused by the acidification of the growth medium (Fig. 2B). Similar results as with CGXII medium were obtained for MMI minimal medium (data not shown). Attempts to identify the substance responsible for acidification revealed that neither l-glutamate, which is often produced in C. glutamicum strains with an altered cell wall (Eggeling and Sahm, 1999), nor lactate, which is sometimes formed by C. glutamicum under non-optimal growth conditions, was present in significant concentrations in the supernatant of ΔmtrAB cultures. Thus, the identity of the acid formed by the ΔmtrAB strain remains unknown at present.
Morphology and antibiotics susceptibility of the C. glutamicum ΔmtrAB mutant
In the routine analysis of the ΔmtrAB mutant by light microscopy, it became obvious that the morphology of ΔmtrAB cells differed significantly from that of wild-type cells, in particular concerning the cell length. These observations were confirmed by the scanning electron micrographs shown in Fig. 3. Whereas C. glutamicum wild-type cells had an average length of 1.43 µm, ΔmtrAB cells were almost three times as long (3.75 µm on average). Furthermore, the mutant but not the wild-type cells were segmented, with the individual segments differing in diameter. Further support for the abnormal cell morphology of the ΔmtrAB cells was obtained by transmission electron microscopy as shown in Fig. 4. Some of the mutant cells clearly showed an irregular septum formation.
The morphological changes indicated that the lack of the MtrAB two-component system led to changes in cell wall metabolism. In order to corroborate further such changes, the sensitivity of the ΔmtrAB mutant and the wild type to inhibitors of cell wall synthesis was tested. Preliminary experiments with L plates containing the selected antibiotics revealed that the mutant is more sensitive to vancomycin and lysozyme, but more resistant towards ethambutol (data not shown). Minimal inhibitory concentrations (MICs) were subsequently determined for penicillin, vancomycin and ethambutol using Etest® stripes (AB Biodisk). Penicillin and vancomycin inhibit the transpeptidation reaction by inhibiting the transpeptidase enzymes and by binding to the terminal D-Ala-D-Ala residues of uncross-linked peptides present in the peptidoglycan network and in lipid II (Hubbard and Walsh, 2003) respectively. In contrast to these inhibitors of peptidoglycan synthesis, ethambutol interferes with the synthesis of the arabinogalactan moiety of the cell wall by inhibition of membrane-bound arabinosyl transferases (Belanger et al., 1996). As shown in Table 1, the ΔmtrAB mutant cells were more sensitive to penicillin and vancomycin, but much more resistant to ethambutol. The latter observation is of high interest from a medical point of view, because in about 25% of clinically isolated ethambutol-resistant M. tuberculosis strains the molecular basis of resistance is unknown (Ramaswamy et al., 2000).
Table 1. Minimal inhibitory concentrations (MICs) of penicillin G, vancomycin and ethambutol for growth of different C. glutamicum strains.
C. glutamicum strain
MIC (µg ml−1)
The MICs were determined with Etest® stripes as described in Experimental procedures. Mean values and standard deviations were determined from at least three independent experiments.
0.50 ± 0.14
0.91 ± 0.30
3.21 ± 1.58
0.32 ± 0.07
0.89 ± 0.32
1.57 ± 0.35
0.06 ± 0.01
0.43 ± 0.16
0.52 ± 0.21
0.86 ± 0.31
5.00 ± 2.29
Comparative transcriptome analysis of wild type and strainΔmtrAB
The pleiotropic phenotype of the ΔmtrAB mutant strain showed that the MtrAB two-component system affects C. glutamicum physiology under the tested culture conditions. As MtrA acts most probably as a transcriptional regulator, it was reasonable to assume that the role of this two-component system might be elucidated by comparing the transcriptional profiles of the C. glutamicum wild type and the ΔmtrAB mutant under the same culture conditions. Whole-genome DNA microarrays containing PCR probes representing 3541 C. glutamicum open reading frames (ORFs) (Lange et al., 2003; Wendisch, 2003) were used to find genes which display altered mRNA levels in the mutant strain. Table 2 lists those genes whose average mRNA level determined in three independent experiments differed at least threefold in the two strains (see also TableS1). The RNA was isolated from cells grown in CGXII minimal medium to an OD600 of ≈ 5, where both strains still had the same growth rate and the pH of the ΔmtrAB culture was still at ≈ pH 6.5.
Table 2. Results of three DNA microarray experiments including standard deviation and P-value starting from three independent cultures of C. glutamicum wild type and the ΔmtrAB mutant grown in CGXII minimal medium with 2% (w/v) glucose.
Except for mtrB, only genes were included whose mRNA level differed at least threefold in the two isogenic strains and where the P-value was below 0.05.
14.51 ± 8.90 (0.01)
Lipoprotein with unknown function
4.45 ± 2.19 (0.03)
Membrane-bound protease modulator
3.90 ± 0.23 (0.01)
Proline/ectoine uptake carrier
0.12 ± 0.06 (0.01)
Betaine uptake carrier
0.17 ± 0.08 (0.02)
0.27 ± 0.15 (0.02)
Sensor histidine kinase
0.45 ± 0.11 (0.01)
Hypothetical protein (IS element)
0.30 ± 0.14 (0.02)
Hypothetical protein (IS element)
0.31 ± 0.16 (0.04)
Hypothetical membrane protein
0.32 ± 0.21 (0.04)
0.33 ± 0.08 (0.01)
The mRNA level of three genes was more than threefold increased in the ΔmtrAB mutant, i.e. NCgl2411, NCgl2737 and lpqB. In silico analysis revealed that the NCgl2411 and NCgl2737 proteins possess conserved domains implicated in proteolytic processes. NCgl2411 (191 residues) contains in its carboxy-terminal half a metallopeptidase domain of the M23B family within the merops protease database (Rawlings et al., 2002). Members of this family contain the characteristic motif HXH, the histidine residues of which presumably serve as zinc ligands. Therefore, the NCgl2411 gene was named mepA for metallopeptidase A. The MepA protein contains three hydrophobic regions extending from amino acid residues 14–34, 48–68 and 94–114. The first one maybe part of a putative signal peptide with the most likely cleavage site between residues 34 and 35 (ASA↓QT). Irrespective of whether MepA is a soluble or an integral membrane protein, the metallopeptidase domain is most probably located extracytoplasmatically. The NCgl2737 protein (325 residues) contains in its carboxy-terminal part (residues 94–275) an SPFH domain, named after the initials of the included protein families, i.e. stomatins, prohibitins, flotillins and HflK/C (Tavernarakis et al., 1999). The characterized members are membrane-bound proteins that modulate the activity of proteases, e.g. the periplasmic domain of the E. coli HflK/C complex has been shown to modify the proteolytic activity of FtsH (Kihara et al., 1996). The NCgl2737 protein is presumably also membrane-bound because it contains two putative transmembrane helices extending from residues 38–58 and from 73–93. In the following, the NCgl2737 gene was named ppmA for putative protease modulator. A third gene with a more than threefold elevated mRNA level in the ΔmtrAB mutant was lpqB, the gene located immediately downstream of mtrB (Fig. 1). This indicates that lpqB and possibly also mtrAB might be negatively regulated by the MtrAB system. The mRNA ratio of the genes sahH and tmk located upstream of mtrA (Fig. 1) were 0.9 and 0.8, respectively, suggesting that they are not coregulated with lpqB.
The mRNA level of six genes (not including mtrA and mtrB) was more than threefold decreased in the ΔmtrAB mutant (Table 2), four of which encode hypothetical proteins. The genes with the strongest reduced RNA level were betP and proP, which encode secondary transporters for the uptake of the compatible solutes betaine and proline/ectoine respectively (Peter et al., 1996; 1998a). Thus, the absence of the mtrAB genes has a marked influence on genes involved in osmoprotection.
The mRNA ratios (ΔmtrAB/wild type) of mtrA and mtrB itself were 0.27 and 0.45 respectively (Table 2). Because both genes were almost completely deleted from the chromosome, much lower ratios had been expected. The high ratios actually found most likely result from cross-hybridization with mRNA of genes encoding one of the other 12 sensor kinases and response regulators of C. glutamicum.
Verification of the DNA microarray results by primer extension and 5′-RACE analyses
In order to independently verify the altered mRNA levels of the genes mepA (NCgl2411), ppmA (NCgl2737), betP and proP in the ΔmtrAB mutant, primer extension analyses were performed for these genes using two different oligonucleotides per gene. For mepA and ppmA, primer extension products were only observed in the ΔmtrAB mutant, whereas in the case of betP and proP primer extension products were only visible in the wild type (Fig. 5). These results were in accordance with the DNA microarray studies. In the case of mepA, three successive bases (ACA) were identified as transcriptional start sites, which were located 69–67 bp upstream of the mepA start codon. In the case of ppmA, transcription was found to start at a single C base located 43 bp upstream of the putative start codon. For proP, three successive bases (TCA) were identified as start points, located 142–144 bp upstream of the start codon. In the case of betP, two primer extension products were identified ending at two successive T bases immediately upstream of the ATG start codon. In all cases, the same transcriptional start sites were determined with a second primer annealing at a different position (data not shown). In the case of proP and betP, the transcriptional start sites were confirmed independently by means of the 5′-RACE technique using wild-type RNA isolated before and 60 min after a hyperosmotic shock (data not shown). The transcriptional start points were identical independent of the growth conditions, indicating that transcription of proP and betP is dependent on a single promoter in each case. In Fig. 6, an alignment of the promoter regions of mepA, ppmA, betP and proP is shown.
Effects of mepA overexpression in E. coli and C. glutamicum
As outlined before, the ΔmtrAB mutant has an altered cell morphology and antibiotics susceptibility. A survey of the genes with changed mRNA level in the mutant suggested that the strongly increased expression of mepA could be primarily responsible for this phenotype, because the encoded protein might function as ‘periplasmic’ metallopeptidase, whose target could be the murein sacculus. This idea was supported by the finding that overexpression of the E. coli lipoprotein NlpD, which also contains the M23B metallopeptidase domain as well as a LysM domain predicted to be involved in peptidoglycan binding, resulted in an irregular cell morphology (Lange and Hengge-Aronis, 1994). Therefore, it was tested whether overexpression of the C. glutamicum mepA gene triggered morphological changes, too. For this purpose, the expression plasmid pEKEx2-mepA was constructed, which contains the mepA gene under control of an IPTG-inducible tac promoter. Introduction of this plasmid into E. coli DH5α led to a significant lysis of the cells when they were cultivated in Luria–Bertani (LB) medium even in the absence of IPTG. Resuspension of these cells in water led to an immediate and almost complete lysis of the cells (data not shown), suggesting severe damage of cell wall integrity and a resulting inability to cope with hypo-osmotic stress. As shown in Fig. 7B, expression of the C. glutamicum mepA gene in E. coli with plasmid pEKEx2-mepA led to a significant elongation of the cells, i.e. a similar phenotype as observed with the C. glutamicumΔmtrAB mutant. In the control strain E. coli DH5α transformed with pEKEx2, no change in cell morphology was observed (Fig. 7A). The effect of the C. glutamicum mepA gene on the E. coli cells might be explained by the assumption that MepA damages the peptidoglycan part of the cell wall.
In contrast to E. coli, transformation of C. glutamicum wild type with pEKEx2-mepA had no obvious effect on the cell morphology, independent of the presence or absence of IPTG (data not shown).
Altered expression of genes involved in osmoprotection in the ΔmtrAB mutant
For this purpose, wild-type and ΔmtrAB cells were cultivated in MMI medium. When they had reached the mid-exponential growth phase, a severe osmotic upshock was applied by the addition of 1 M NaCl resulting in an increase of the osmolality from 0.3 to 2.2 osM. Cells were harvested before (T0) and at 5–180 min (T5–T180) after the shock and the total RNA was isolated. The relative mRNA levels of the abovementioned genes were detected by means of RNA hybridization experiments. These analyses showed that the absence of mtrAB has a strong influence on the expression of betP, lcoP and proP(Fig. 8A). In wild-type cells, the corresponding transcripts were detectable already before the shock and significantly increased within 30–60 min after the upshock by a factor ranging between 3 and 60 respectively. In the mtrAB mutant, betP and lcoP transcripts were barely detectable before or after the osmotic shock. In the case of proP, low transcript levels were observed from 60 to 180 min after the shock. Remarkably, the ectP gene encoding an ectoine/betaine/proline uptake system was induced both in the wild type and the mutant to comparable levels, showing that its expression was not influenced by the mtrAB deletion. The analysis of mscL and yggB, which encode mechanosensitive channels, revealed that the expression profile of yggB was similar in wild type and the mutant, whereas the pattern of the mscL transcript was different in the two strains (Fig. 8B). In the ΔmtrAB mutant the expression of mscL was elevated even under low osmolality conditions. These amounts were elevated by a factor of ≈ 3–4 up to 30 min after the shock but then reached levels similar to wild type. The expression profile of genes encoding enzymes for the biosynthesis of compatible solutes were not changed in the ΔmtrAB mutant compared to the wild type (Fig. 8C). In both strains proB, otsA and treS were significantly and treY marginally induced with maximal mRNA levels 30 min after the osmotic upshock.
The expression changes concerning the genes coding for uptake systems of compatible solutes betP, lcoP and proP led to the question whether the mutant strain has a growth disadvantage compared to wild-type cells if hyperosmotic conditions were applied. For this purpose exponentially growing ΔmtrAB and wild-type cells which were adapted to MMI minimal medium were exposed to a hyperosmotic shift by the addition of 1 M NaCl either in the presence or in the absence of 10 mM betaine. After the upshock, both strains had identical growth rates of 0.16–0.17 h−1 irrespective of the absence or presence of betaine. This shows that the mutant has no growth disadvantage under the conditions tested. In a previous study (Peter et al., 1998a), where wild-type cells of an overnight culture were directly inoculated into high osmolality medium, the presence of betaine had a positive effect on the growth rate. By contrast, in the present study, where exponentially growing cells were subjected to a hyperosmotic shift, betaine had no effect.
To investigate whether the changed cell morphology of the ΔmtrAB strain led to increased cell lysis after a downshock, the ability of wild-type and mutant cells to cope with hypo-osmotic conditions was tested. For this purpose cells were first adapted to an osmolality of 2.4 osmol kg−1 before they were transferred into minimal medium with an osmolality of 0.9 osmol kg−1. The ratio of cells with intact membranes versus damaged membranes was used as an indicator for cell lysis as recently described (Nottebrock et al., 2003). In the case of the wild type, 76 ± 14% of the cells stayed alive, in the case of the ΔmtrAB mutant 86 ± 22%, indicating that the elongated mutant cells are as robust as wild-type cells concerning the adaptation to an osmotic downshock. This result might be explained by the increased expression of mscL in this strain.
In order to test whether additional two-component systems of C. glutamicum are involved in the regulation of the genes involved in osmoprotection, similar experiments as described in Fig. 8 were carried out with 11 strains each carrying a deletion of the genes for one specific two-component system (S. Schaffer and M. Bott, unpubl.). The dot blot analyses showed that the absence of these other systems had no influence on the expression of the indicated genes.
The manifold consequences of the mtrAB deletion raised the question whether reintroduction of the mtrAB genes into the ΔmtrAB mutant could restore the wild-type characteristics. To this end, plasmid pEKEx2-mtrAB was constructed in which expression of the mtrAB genes is controlled by the IPTG-inducible tac promoter. This plasmid and the vector pEKEx2 were transferred by electroporation into the ΔmtrAB strain and into the wild type. Growth experiments in CGXII minimal medium with 4% glucose revealed that the acidification of the medium by the ΔmtrAB mutant (see Fig. 2B) could not be abolished by pEKEx2-mtrAB, irrespective of the presence (1 mM) or absence of IPTG. By contrast, the morphology of the cells was clearly reverted to a wild type-like morphology in the presence of pEKEx2-mtrAB and IPTG, as shown in Fig. 9. Similarly, the antibiotics susceptibility of the ΔmtrAB strain carrying pEKEx2-mtrAB was like that of the wild type when the cells were grown in the presence of IPTG (Table 1).
In order to test whether the expression changes in the ΔmtrAB strain could be complemented by plasmid-encoded expression of mtrAB, the RNA dot blot experiment shown in Fig. 8 was repeated using wild type/pEKEx2 and ΔmtrAB/pEKEx2 as reference strains and ΔmtrAB/pEKEx2-mtrAB as complementation strain. For induction of the plasmid-encoded mtrAB genes, 0.2 mM IPTG was present during cell cultivation. As shown in Fig. 10A, the mRNA pattern of betP, proP, lcoP and mscL in strain ΔmtrAB/pEKEx2-mtrAB was identical to that of the wild type. Expression of the cell wall-related genes, mepA, ppmA and lpqB, was investigated in a similar way (Fig. 10B). For this purpose, the three abovementioned strains were grown in minimal medium in the presence of 50 µg ml−1 kanamycin and 0.2 mM IPTG. Cells were harvested in the early exponential phase (OD600 6), as in the DNA microarray experiments, and in the late exponential phase (OD600 15). The RNA hybridization experiment revealed that the elevated mRNA level of mepA in the mtrAB deletion mutant was reduced to the wild-type level in the complementation strain. In contrast, the increased mRNA level of lpqB of the ΔmtrAB mutant was not reverted to wild-type levels in the complementation strain. This result might be related to the localization of lpqB immediately downstream of the mtrAB genes, which are almost entirely deleted in the complementation strain. If these genes form a transcriptional unit, the altered mRNA structure caused by the deletion might have prevented a successful complementation. In case of ppmA, the hybridization signals did not allow a conclusive evaluation (data not shown).
In summary, expression of the plasmid-encoded mtrAB genes in the ΔmtrAB mutant could reverse the changes in morphology and antibiotics susceptibility as well as the changes in the mRNA levels of proP, betP, lcoP, mscL, and mepA. No complementation was achieved with respect to the lpqB mRNA levels and with respect to the acidification of the medium. The latter phenotype might thus result from a mutational event that occurred during or as a consequence of the mtrAB deletion.
The two-component signal transduction system consisting of the sensor histidine kinase MtrB and the response regulator MtrA is highly conserved in all known species of Corynebacterium and Mycobacterium, including M. leprae. This indicates that it plays an important role in the physiology of these organisms. In this work, first hints on the function of MtrAB were obtained by the analysis of a ΔmtrAB mutant of C. glutamicum. The ability to delete the mtrA gene of C. glutamicum was in contrast to the studies with M. tuberculosis, where this was not possible (Zahrt and Deretic, 2000). An obvious explanation for this difference cannot be given presently. A comparison of the C. glutamicumΔmtrAB mutant with the isogenic wild-type strain provided evidence for the involvement of the MtrAB system in cell wall metabolism and osmoregulation. Evidences for a role in cell wall metabolism were the radically different morphology of the mutant cells (Fig. 3) and their altered response towards inhibitors of cell wall synthesis (Table 1). The mutant cells were more sensitive to penicillin, vancomycin and lysozyme, indicating a disturbed peptidoglycan sacculus, but on the other hand were more resistant to ethambutol. Ethambutol is an anti-tuberculous drug that inhibits membrane-bound arabinosyltransferases (encoded by the emb genes) involved in the synthesis of the arabinogalactan moiety present in the cell walls of mycobacteria and corynebacteria (Belanger et al., 1996). It is known that increased expression of the emb genes (Belanger et al., 1996) and certain amino acid exchanges in the Emb proteins and in a limited number of other target proteins can confer increased ethambutol resistance (Ramaswamy et al., 2000). However, there is also clear evidence for the presence of additional, yet unknown targets responsible for ethambutol resistance (Ramaswamy et al., 2000). In the genome of C. glutamicum, a single emb gene encoding a membrane-bound arabinosyltransferase is present (NCgl0184). Whereas mutations in the emb gene of the C. glutamicumΔmtrAB mutant cannot be ruled out, an increased emb mRNA level was not observed by the DNA microarray experiments. Thus, it is feasible that one or several of the other detected differences in gene expression might be related to the enhanced ethambutol resistance of strain ΔmtrAB.
Comparison of the transcriptome of wild type and ΔmtrAB mutant by whole-genome DNA microarrays revealed a limited number of genes whose mRNA level differed at least threefold in the two strains (Table 2). Therefore, expression of these genes could be directly or indirectly influenced by the MtrAB system. The most drastic difference was obtained for the mepA gene (NCgl2411), whose mRNA level was 14-fold higher in the ΔmtrAB mutant than in the wild type, indicating that its expression might be repressed by MtrAB. It encodes a protein with a presumably extracytoplasmic metallopeptidase domain of the M23B family (Rawlings et al., 2002), whose target could be the peptides already incorporated into the peptidoglycan sacculus or lipid II peptides that will be incorporated. This would offer an explanation for the increased sensitivity of the ΔmtrAB mutant to lysozyme, penicillin and vancomycin. Support for the assumption that the MepA protein targets peptidoglycan came from the observation that expression of the mepA gene in E. coli led to elongated cells that were hypersensitive to an osmotic downshift. Almost all cells were lysed when they were transfered from LB medium to water. In this context, it is noteworthy that C. glutamicum and E. coli contain the same peptidoglycan type A1γ (Schleifer and Kandler, 1972). Overexpression of mepA in C. glutamicum wild type, however, did not cause an altered morphology, indicating that additional changes besides the increased mepA mRNA level are necessary to produce the ΔmtrAB phenotype. A second gene with a more than threefold increased mRNA level in the ΔmtrAB mutant was ppmA (NCgl2737), which encodes a membrane-bound protein with an SPFH domain (Tavernarakis et al., 1999). Characterized proteins containing such a domain were reported to modulate the activity of proteases, e.g. HflK/C of E. coli modulates the activity of the protease FtsH (Kihara et al., 1996). A possible target of the PpmA protein might be the metalloprotease MepA.
A second group of genes whose expression was strongly influenced by the mtrAB deletion encodes proteins involved in osmoprotection. In RNA dot blot experiments, the mRNA levels of betP, proP and lcoP, all of which encode secondary carriers for the uptake of compatible solutes, were strongly decreased in the mutant, both before and after a hyperosmotic shock. In contrast, the mRNA level of the mscL gene encoding a mechanosensitive channel (Nottebrock et al., 2003) was increased in the mutant before the hyperosmotic shift. In the microarray experiments, the transcript ratio (ΔmtrAB/wild type) for betP and proP was the lowest of all genes (>0.2), that for lcoP was not analysable and that for mscL was 1.8, i.e. increased, but below our cut-off value. Whereas BetP, ProP and LcoP are involved in the response to hyperosmotic stress, the mechanosensitive channel MscL is required to cope with hypo-osmotic stress. An inverse transcriptional regulation of these genes therefore seems reasonable. In contrast to the four above-mentioned genes, the mRNA levels of ectP, yggB, otsA, proB, treS and treY were unchanged in the ΔmtrAB mutant both in the dot blots and the microarray experiments, where mRNA ratios (ΔmtrAB/wild type) between 0.6 and 1.3 were determined for them. They encode a secondary carrier for uptake of compatible solutes (EctP), a mechanosensitive channel of small conductance (YggB), and enzymes involved in the synthesis of the compatible solutes trehalose (OtsA, TreS, TreY) and proline (ProB). It is remarkable that exclusively the synthesis of carriers and channels involved in osmoprotection is influenced by the mtrAB deletion, but not the synthesis of enzymes required for the endogenous formation of compatible solutes. An important aspect regarding BetP and ProP is that these carriers regulate their own activity in response to osmolality, with negligible activity at low osmolality and strong activity at high osmolality (Peter et al., 1996; 1998a; Rübenhagen et al., 2000; 2001). In the case of BetP, the amino- and carboxy-terminal extensions were shown to be critical for activity regulation (Peter et al., 1998b). The osmosensory and osmoregulatory properties of BetP and ProP allow for a rapid activation of these carriers in response to hyperosmotic stress, whereas the transcriptional regulation of their genes might be considered as a long-term adaptation process. The result that the ΔmtrAB mutant cells had no disadvantage compared to the wild type after a hyperosmotic shock in the absence of betaine can be explained by the fact that expression of the genes responsible for the synthesis of compatible solutes is not altered by the mtrAB deletion. However, even in the presence of 10 mM betaine the mutant grew like the wild type after an osmotic upshift, indicating that the activity of residual betaine carrier, EctP, is sufficient to cope with this stress. In the response to a hypo-osmotic shift, the ΔmtrAB mutant appeared to be even more resistant than the wild-type, which might be a consequence of the increased expression of the mscL gene.
Considering that MtrA most probably acts as a transcriptional regulator and that the genes whose expression is most strongly influenced by the mtrAB deletion are involved in peptidoglycan metabolism and osmoprotection, several scenarios can be derived: (i) MtrAB is directly involved in the regulation of cell wall synthesis, in particular via repression of mepA, and the altered expression of genes involved in osmoprotection is an indirect effect, (ii) MtrAB is directly involved in osmoregulation and the effect on the expression of mepA, ppmA and lpqB is indirect and (iii) MtrAB is directly involved both in the regulation of cell wall synthesis and in osmoregulation. To determine which of these or other possible models reflects the in vivo function of the MtrAB two-component system requires knowledge of the direct target genes of the response regulator MtrA. Therefore, our future studies will focus on the identification of these primary target genes using assays based on MtrA–DNA interaction.
In the past, a number of two-component systems involved in either osmoprotection or cell wall biosynthesis were described, but few data are available on systems influencing both of these processes. One recent example is the VraSR two-component system of Staphylococcus aureus, which was proposed to be a positive regulator of cell wall peptidoglycan biosynthesis (Kuroda et al., 2003). Comparative transcriptome analysis of a ΔvraSR mutant with the wild type revealed that the gene with the most strongly decreased mRNA level in the mutant was proP, encoding a proline/betaine transporter. Evidence was obtained that the VraSR system is activated by vancomycin treatment of the cells, because many genes whose expression was induced in the wild type upon exposure to vancomycin were no longer induced in the ΔvraSR mutant, including several genes involved in cell wall biogenesis. However, as in the case of the MtrAB system, the direct target genes of the VraSR system are still unknown.
Bacterial strains, media and growth conditions
Bacterial strains and plasmids used or constructed in the course of this work are listed in Table 3, oligonucleotides in Table 4. C. glutamicum was cultivated aerobically on a rotary shaker (150 r.p.m.) at 30°C in LB medium (Sambrook et al., 1989), CGXII minimal medium (Keilhauer et al., 1993) or MM1 minimal medium (Nottebrock et al., 2003). CGXII and MM1 medium contained 4% (w/v) glucose as carbon source. For strain construction and maintenance, BHIS agar plates [BHI agar (Difco) with 0.5 M sorbitol] were used. E. coli DH5α was grown aerobically on a rotary shaker (150 r.p.m.) at 37°C in LB medium or plated onto LB agar plates [LB medium with 1.5% (w/v) agar]. If appropriate, kanamycin was added to final concentrations of 25 µg ml−1 (C. glutamicum) or 50 µg ml−1 (E. coli). High osmolality was adjusted by the addition of 1.05 M NaCl (equivalent to an osmotic upshift of ≈ 2 osM) and checked by means of freezing point reduction with an Osmomat 030 (Gonotec GmbH).
. In some cases oligonucleotides were designed to introduce recognition sites for restriction endonucleases (recognition sites underlined, restriction endonucleases indicated in parentheses) or complementary 21mer sequences for generating cross-over PCR products (in italics).
Construction and verification of the ΔmtrAB mutant
CCC GGG TAA TAC GAC TCA CTA TAG GGC GTT CGT TCC GCA TTC G
GTG AGT TTT ATG ACT ACA TCT GAC
GCT TCA TCA ATG CGG CC
GCA AAC ATC ATG GAA ATC
GGT TCT GTG GTT CCG TAG
GAA GGC AGT ACC GAT GAC
CAA CAC CCT TGC GCT TTG
GTT CTG CTT CGA TGG AAG C
GCT CCT GCA GGA GAT CG
GAA CAA CGT TGA CGT CAT C
CAG CAA CAT AGG TGG TCG C
AAG CTC CGA AGA AAC CTC
GCA AAG CCC AGG TAG GAA C
Cells for the isolation of RNA for hybridization experiments and 5′-RACE analyses were first cultivated overnight in BHI. These precultures were washed in 0.9% NaCl solution and used for inoculation of MMI medium (0.3 osM). In the mid-exponential growth phase (OD600 6–7), the cells were harvested and resuspended in MMI medium with 1.05 M NaCl to exert a sudden osmotic upshift. Before and at different times after the osmotic upshift, cells from 2 ml of culture were harvested and used for RNA isolation.
General DNA techniques and sequence analyses
Standard methods like PCR, restriction or ligation were carried out according to Sambrook et al. (1989). E. coli was transformed using standard methods (Inoue et al., 1990). DNA sequencing was performed either with an ABI 310 sequencer (Perkin Elmer) or with a LI-COR 4200 sequencer (Licor Inc.). Sequencing reactions were carried out with the ABI PRISMTM Big Dye Terminator Cycle Sequencing Kit (Perkin Elmer) or with the Thermosequenase fluorescent sequencing kit (Amersham Pharmacia Biotech).
Mutant and plasmid construction
An in-frame mtrAB deletion mutant of C. glutamicum was constructed via a two-step homologous recombination procedure as described previously (Niebisch and Bott, 2001). The upstream region of mtrA and the downstream region of mtrB (≈ 500 bp each) were amplified using the oligonucleotide pairs Delta-mtrAB-1/Delta-mtrAB-2 and Delta-mtrAB-3/Delta-mtrAB-4, followed by a cross-over PCR with oligonucleotides Delta-mtrAB-1 and Delta-mtrAB-4. The resulting PCR product of about 1 kb was digested with XmaI and XbaI and cloned into pK19mobsacB (Schäfer et al., 1994) cut with the same enzymes. Transformation of the resulting plasmid pK19ΔmtrAB into C. glutamicum and screening for the first and second recombination event were performed as described previously (Niebisch and Bott, 2001). The chromosomal deletion was confirmed by PCR with the oligonucleotides Delta-mtrAB-out-fw and Delta-mtrAB-out-rev and Southern blot analysis. The latter was performed as described previously (Niebisch and Bott, 2001).
For IPTG-inducible expression of the C. glutamicum mepA gene, the coding region and 64 bp upstream DNA were amplified with the oligonucleotide pair NCgl2411-pEKEx2-SacI/NCgl2411-pEKEx2-EcoRI-r and the Expand High Fidelity PCR system (Roche Diagnostics). The PCR product of 664 bp was digested with SacI and EcoRI and cloned into pEKEx2 (Eikmanns et al., 1991) cut with the same enzymes. The resulting plasmid was named pEKEx2-mepA.
For IPTG-inducible expression of mtrAB, the coding regions and 32 bp upstream DNA were amplified with the oligonucleotides MtrA_BamHI_fw and MtrB_BamHI_rv, which both contained BamHI restriction sites, and PfuTurbo Cx Hotstart DNA polymerase. After addition of adenosine overhangs with Taq DNA polymerase, the PCR product of 2.3 kb was cloned into pCR2.1-TOPO (Invitrogen). After digestion of pCR2.1-mtrAB with BamHI, the mtrAB fragment was cloned into pEKEx2 cut with the same enzyme and dephosphorylated. In the resulting plasmid pEKEx2-mtrAB, expression of the mtrAB genes is controlled by the tac promoter. DNA sequence analyses of pEKEx2-mtrAB plasmids from three different clones revealed that none of the cloned mtrAB fragments contained deviations from the published wild-type sequence.
Preparation of total RNA for DNA microarray and primer extension experiments
Cultures of the wild type and the ΔmtrAB mutant were grown in CGXII minimal medium containing 2% (w/v) glucose. At an OD600 of 5–6, cells were in the exponential growth phase, both strains showed comparable growth rates (0.28–0.33 h−1) and the pH of the ΔmtrAB culture was still above 6.5. At this time point, 20 ml of the cultures were poured into ice-containing tubes pre-cooled to −20°C and cells were harvested by centrifugation (3 min, 4200 g, 4°C). The cell pellet was either directly used for RNA isolation or frozen in liquid nitrogen and stored at −70°C. For isolation of RNA, the (frozen) cell pellet was resuspended in 350 µl of RLT buffer of the RNeasy kit (Qiagen). Then, 250 mg of 0.1 mm diameter zirconium-silica beads (Roth) were added and the cells were disrupted by 30 s of bead beating with a Silamat S5 (Vivadent). After centrifugation (1 min, 14 000 g), the supernatant was used for RNA preparation by using the RNeasy system (Qiagen) with on-column DNase I treatment according to the manufacturer's instructions. Isolated RNA samples were analysed for quantity by UV spectrophotometry and stored at −20°C until use.
DNA microarray analyses
The generation of whole-genome DNA microarrays was performed as described previously (Lange et al., 2003). Identical amounts (20–25 µg) of total RNA were used for random hexamer-primed synthesis of fluorescently labelled cDNA by reverse transcription with Superscript II (Invitrogen) and the fluorescent nucleotide analogues FluoroLink Cy3-dUTP (green) and Cy5-dUTP (red) (Amersham Pharmacia) as described previously (Wendisch et al., 2001). The labelled cDNA probes were purified and concentrated by using Microcon YM-30 filter units (Millipore). Subsequently, the mixed Cy3- and Cy5-labelled cDNAs containing 1.2 µg poly(A) µl−1 (Sigma) as competitor, 30 mM Hepes and 0.3% sodium dodecyl sulphate in 3× SSC were hybridized to whole-genome arrays in a humid chamber for 5–16 h at 65°C. After hybridization, the arrays were washed in 1× SSC containing 0.03% sodium dodecyl sulphate and finally in 0.05× SSC. Immediately after stringent washing, the fluorescence intensities at 635 and 532 nm were determined using a GenePix 4000 laser scanner (Axon Inc.) and the TIFF images were processed by using GenePix 3.0 software. Data were normalized to the average fluorescence ratio for C. glutamicum genomic DNA (represented by up to 200 spots per DNA microarray), which was defined as one. The normalized ratio of median was taken to reflect the relative RNA abundance for spots with a green or red fluorescent signal that was at least threefold greater than the median fluorescence background signal. For statistical analysis of the gene expression data, P-values for the independent replicate experiments were calculated based on the Student's t-test by using log-transformed fluorescence ratios for individual genes on the one hand and for genomic DNA on the other hand. A heteroscedastic t-test was performed using the corresponding function of the Microsoft Excel software. Only genes with P-values < 0.05 were considered to show significantly changed RNA levels. In total, three competitive hybridizations were carried out, each starting with independent cultures of wild type and ΔmtrAB mutant.
Primer extension analysis
Non-radioactive primer extension analysis was performed as described previously (Engels et al., 2004). Briefly, 10 µg of total RNA were combined with 2 pmol IRD800-labelled oligonucleotide (MWG Biotech) and 5× annealing buffer (50 mM Tris-HCl pH 7.9, 1.25 M KCl) in a total volume of 10 µl. After incubation at 65°C for 5 min, samples were slowly (0.5°C per 2 min) cooled down to 42°C and reverse transcription was initiated by adding a mix consisting of 23 µl of water, 10 µl of 5× reverse transcription buffer (250 mM Tris-HCl pH 8.3, 125 mM KCl, 15 mM MgCl2), 5 µl of 100 mM DTT, 1 µl of deoxyribonucleotides (25 mM dATP, dCTP, dGTP and dTTP), 0.5 µl of actinomycin D (5 mg ml−1 in ethanol) and 0.5 µl of SuperScript III RNase H– Reverse Transcriptase (200 units µl−1; Life Technologies) and allowed to proceed for 1 h at 42°C. Subsequently, RNA was degraded by adding 120 µl of RNase A reaction mix (10 mM Tris-HCl pH 7.5, 1 mM EDTA, 100 mM NaCl, 100 µg ml−1 sonicated salmon sperm DNA, 200 µg ml−1 RNase A, added freshly) and incubation for 1 h at 37°C. cDNA was precipitated overnight and dissolved in 2 µl of water and 2 µl of formamide loading dye (Epicentre Technologies Corp.). One microlitre was then loaded onto a denaturing 4.6% (w/v) Long Ranger (Biozym) sequencing gel (separation length 42 cm) and separated in a Long Read IR DNA sequencer (Licor Inc.). The length of the reaction products was determined by running the four lanes of a DNA sequencing reaction set-up using the same oligonucleotide as for reverse transcription alongside the primer extension products. For DNA sequencing, PCR products of the respective promoter regions were used as template. Each transcriptional start was determined using at least two different oligonucleotides (NCgl2411-PE-1 and NCgl2411-PE-2 for mepA; 225-PE-1 and 225-PE-2 for ppmA; betP-PE-1, betP-PE-2 and betP-PE-3 for betP; proP-PE-1, proP-PE-2 and proP-PE-3 for proP).
RNA dot blot experiments
For purification of RNA from C. glutamicum, cells harvested from 2 ml of culture were sedimented by centrifugation at 30°C, suspended in buffer RA I (Macherey-Nagel) and broken with 300 mg glass beads (0.1–0.25 mm diameter) by two sequential passages of 30 s and 6.5 m s−1 in a FastPrep®120 instrument (Q-Biogene). RNA was purified from the disrupted cells with the Nucleospin®RNAII kit (Macherey-Nagel) according to the manufacturer's instructions. For the construction of RNA anti-sense probes of the desired gene (cf. Table 2) intragenic fragments of a size of roughly 500 bp were amplified using ATCC13032 cells as template and the oligonucleotides shown in Table 4. The resulting DNA fragments were ligated with pDrive (Qiagen) or pGEM derivatives (Promega) (Table 2). In the case of mepA, lpqB and ppmA, PCR fragments including the T7 promoter rather than plasmids were used for probe generation. DIG-11-dUTP-labelled anti-sense RNA was obtained by in vitro transcription (1 h, 37°C) from linearized vectors using T7 RNA polymerase. Changes in gene transcription were monitored by RNA hybridization experiments using these digoxygenin-labelled anti-sense RNA probes. For that purpose, ≈ 6 µg of total RNA was transferred to a nylon membrane using a Minifold Dot Blotter (Schleicher and Schuell). RNA was bound to the membrane by careful vacuum suction (15 mbar). RNA was cross-linked to the membrane by means of ultraviolet irradiation at 125 J cm−2. Hybridization and detection steps were carried out according to the DIG application manual (Roche Applied Science). Chemiluminescence was detected with commercially available X-ray films or in case of the densitometric quantification, via the CCD camera of the LAS 1000 CH system (Fuji). The signals were quantified with the program aida image eliza 2.11.
The transcriptional start points were determined by means of the 5′-RACE technique (rapid amplification of 5′-cDNA ends). The 5′,3′-RACE Kit (Roche Applied Science) was used according to the manufacturer's instructions. First strand cDNA was synthesized from total RNA (up to 2 µg) using a gene-specific primer (annealing ≈ 500 bp downstream of the translational start point), reverse transcriptase and dNTPs. The mixture was incubated for 1 h at 55°C and then the single-stranded cDNA was purified from unincorporated primers and nucleotides. Subsequently, terminale transferase was used to add a homopolymeric A-tail to the 3′-end of the single-stranded cDNA, thereby tagging the 5′-end or transcriptional start point of the respective mRNA. Tailed DNA was then amplified by PCR using a oligo-dT-anchor primer (supplied by the manufacturer) and a gene-specific primer (annealing ≈ 480 bp downstream of the translational start point). The obtained cDNA was further re-amplified by a second PCR with a nested gene-specific primer (annealing ≈ 460 bp downstream of the translational start point) and the oligo-dT-anchor primer. The resulting PCR product was used as template in a sequencing reaction to determine the last nucleotide upstream of the poly A-tail, which represents the 5′-end of the respective mRNA (except when the 5′-end starts with one or several A bases).
Culture aliquots were centrifuged for 2 min at 13 000 g and the supernatant was heated for 10–15 min at 80°C in order to inactivate enzymatic activities eventually present. Glucose was determined using the coupled enzymatic assay with hexokinase and glucose 6-phosphate dehydrogenase adapted from Kunst et al. (1984). Each assay mixture consisted of 40 µl of sample (0.1–2.5 mM glucose), 220 µl of 50 mM Tris-maleate buffer pH 6.8 containing 1.13 mM NAD+ and 1.01 mM ATP, 20 µl (1.5 U) of hexokinase and 20 µl (0.5 U) of glucose 6-phosphate dehydrogenase from Leuconostoc mesenteroides (Roche Diagnostics). The latter enzyme accepts NAD+ instead of NADP+ as cofactor. The samples present in a microtiter plate were incubated for 90 min at room temperature with shaking and then absorption was determined at 340 nm using a Thermo-max microtiter plate reader (Molecular Devices). Glucose concentrations were calculated using an extinction coefficient of 6.4 mM−1 cm−1.
Antibiotics susceptibility tests
In the initial tests for antibiotics resistance, the relevant C. glutamicum strains, grown overnight in LB medium with 4% (w/v) glucose, were diluted to an OD600 of 0.1 and one inoculation loop was streaked onto L-plates (10 g l−1 peptone, 5 g l−1 yeast extract, 5 g l−1 NaCl, 1 g l−1 glucose, 15 g l−1 agar) containing the respective antibiotics. For quantitative determination of the antibiotics susceptibility, Etest® stripes containing penicillin, vancomycin or ethambutol were used (AB Biodisk). For this purpose, strains were pre-cultured overnight in 5 ml of LB medium containing kanamycin (25 µg ml−1) and IPTG (1 mM). The next morning 5 ml of the same medium was inoculated with the overnight culture to an OD600 of 0.25 and cultivated at 30°C to an OD600 of about 1.5–2. One hundred microlitres of cell suspension corresponding to an OD600 of 2 were plated onto BHIS plates containing kanamycin (25 µg ml−1) and IPTG (1 mM) and then the Etest stripes were applied. The plates were incubated at 30°C for 19 h, after which the MIC was determined.
Cell viability assay
The viability of cells after an osmotic downshock from 2.4 to 0.9 osmol kg−1 was investigated with the Live/Dead BacLightTM Bacterial Viability Kit (Molecular Probes) as described recently (Nottebrock et al., 2003) using CGXII medium instead of MMI medium.
Scanning electron microscopy
Corynebacterium glutamicum cells prepared for scanning electron microscopy were grown for 18 h at 30°C either on LB or on MMI plates. Afterwards, they were diluted in water to an optical density OD600 of 2. This suspension (10 µl) was added on a polycarbonate filter (filter type 0.2 µm GTTP, 13 mm diameter; Millipore). Subsequently, cells were incubated for at least 18 h in 2.5% glutardialdehyde solution. The filters were washed twice in double-distilled water before cells were successively dehydrated by treatment for 2 h in 20, 40, 60, 80 and 100% acetone respectively. The incubation in 100% acetone was repeated several times to guarantee a complete substitution. For E. coli cells expressing C. glutamicum mepA the procedure was changed as follows because the cells were extremely fragile: cells were cultivated for 16 h in liquid TYGPN medium (Chung et al., 1989), allowing growth even if cells were not shaken. Subsequently, cells were treated for 20 min at 4°C with 1.25% glutardialdehyde and 0.5% osmium tetroxide, which was added directly into the TYGPN medium. After centrifugation and resuspension of cells this incubation was repeated once with a fresh solution. Subsequently, cells were washed three times in double-distilled water before they were diluted to an OD600 between 1 and 8 and 10 µl of these suspensions were added on the polycarbonate filter. The treatment with acetone was performed as mentioned above.
The critical point drying of the samples was carried out in the critical point dryer Polaron (Quorum Technologies). The filters were transferred into the pressure chamber of the critical point dryer which was filled with 100% acetone. After cooling to 7°C the acetone was substituted against liquid CO2. The chamber was filled with liquid CO2 up to five times until the acetone was completely exchanged against CO2, then the system was heated to ≈ 35°C which allows the direct phase transition of CO2 from the liquid into the gas phase. After the drying run, the pressure from the gaseous CO2 was released. Subsequently, cells were coated with gold in a Sputter Coater (Cressington) before being inserted into the SEM (LEO 430i, Zeiss) for observation and imaging.
Transmission electron microscopy
Cells from 5 ml of LB overnight culture were resuspended in 1 ml of 3% glutaraldehyde in 0.1 M sodium phosphate buffer pH 7.2 and fixed for at least 1 h. Subsequently cells were centrifuged for 10 min at low speed and the cell pellet mixed with 3% agarose at 50°C. Blocks of the embedded cells were incubated for 1–4 h in the buffered 3% glutaraldehyde solution and washed overnight in 0.1 M phosphate buffer pH 7.2. Cells were then dehydrated with a standard alcohol series, infiltrated in a mixture of epoxy resin/propylene oxide (1 : 1) for 30 min and flat embedded in pure epoxy resin for 60 min in dried, labelled silicon moulds. The epoxy was allowed to polymerize for 8 h at 37°C and for 56 h at 60°C. Sections (1 µm) were stained with methylene blue/azur II and pre-screened with a light microscope. Ultrathin sections (80 nm) were then prepared with a diamond knife and mounted on copper grids. The grids were stained with uranyl acetate and lead citrate, air-dried and examined with a Philipps transmission electron microscope TEM 400 at 60 kV.
The authors are deeply grateful to Dr Volker Wendisch and his collaborators Dr Tino Polen, Dr Georg Sindelar, Christian Lange, Andreas Krug and Doris Rittmann (Forschungszentrum Jülich) for generously providing whole-genome DNA microarrays of C. glutamicum and for advice on their use. Further on the authors are indebted to Professor Klaus Schmitz (University of Cologne) for support and advice in the preparation of the scanning electron micrographs, Dr Sergey Strelkov (University of Cologne) for GC-MS analysis of culture supernatants and Dr Günther Hollweg (RWTH Aachen University) for performing the transmission electron micrographs. The authors from the Institute for Biotechnology would like to thank Professor Hermann Sahm (Forschungszentrum Jülich) for his continous support.
Table S1. Results of three DNA microarray experiments including standard deviation and P-value starting from three independent cultures of C. glutamicum wild type and the ΔmtrAB mutant grown in CGXII minimal medium with 2% (w/v) glucose.