The MtrAB signal transduction system, which participates in multiple cellular processes related to growth and cell wall homeostasis, is the only two-component system known to be essential in Mycobacterium. In a screen for antibiotic resistance determinants in Mycobacterium smegmatis, we identified a multidrug-sensitive mutant with a transposon insertion in lpqB, the gene located immediately downstream of mtrA–mtrB. The lpqB mutant exhibited increased cell–cell aggregation and severe defects in surface motility and biofilm growth. lpqB cells displayed hyphal growth and polyploidism, reminiscent of the morphology of Streptomyces, a related group of filamentous Actinobacteria. Heterologous expression of M. tuberculosis LpqB restored wild-type characteristics to the lpqB mutant. LpqB interacts with the extracellular domain of MtrB, and influences MtrA phosphorylation and promoter activity of dnaA, an MtrA-regulated gene that affects cell division. Furthermore, in trans expression of the non-phosphorylated, inactive form of MtrA in wild-type M. smegmatis resulted in phenotypes similar to those of lpqB deletion, whereas expression of the constitutively active form of MtrA restored wild-type characteristics to the lpqB mutant. These results support a model in which LpqB, MtrB and MtrA form a three-component system that co-ordinates cytokinetic and cell wall homeostatic processes.
The primary machineries that help bacteria sense and respond precisely to surrounding environments are two-component signal transduction systems. A classical two-component system consists of a transmembrane sensor histidine kinase (HK) and a cytoplasmic response regulator (RR) protein. Upon recognizing certain stimuli, the sensor kinase autophosphorylates one of the histidine residues in its cytoplasmic domain, which then transfers the phosphate group to an aspartate residue of the cognate RR. The phosphorylated RR then becomes active and modulates expression of its regulon (genes regulated by the RR) leading to alterations of cell behaviours in response to the stimuli. The genes encoding the HK and its cognate RR are often genetically linked and therefore classically named ‘two-component’ system.
Accessory proteins of the MtrAB system have not been identified. Based on chromosomal localization, it was speculated that the putative lipoprotein LpqB, whose encoding gene is located immediately downstream of mtrA and mtrB, might act as a modulator of MtrB activity (Hoskisson and Hutchings, 2006). Despite this speculation, the function of LpqB is completely unknown. Here we report the characterization of the first lpqB mutant, which was unexpectedly isolated in a screen for multidrug-sensitive mutants of Mycobacterium smegmatis. Disruption of lpqB resulted in pleiotropic effects including increased multidrug susceptibility, retarded surface motility and biofilm growth, as well as defects in morphological and cell division control. Our experiments showed that LpqB interacts with the extracellular domain of MtrB and affects the MtrA phosphorylation status that mediates alterations in cell division and cell wall homeostasis. Furthermore, lpqB disruption alters expression of dnaA, an MtrA-regulated gene that controls DNA replication. Collectively, the results support a model in which LpqB functions as an accessory protein that modulates activities of the Actinobacterial two-component system MtrAB to provide antibiotic resistance, cell division and cell wall homeostatic control.
Identification of LpqB – a novel determinant of multidrug resistance in M. smegmatis
A library of ∼7000 transposon mutants was generated from wild-type M. smegmatis mc2155 by Himar1-mediated mutagenesis, and deposited in 96-well plates. This library was used to screen for mutants with increased susceptibility to antibiotics (Nguyen et al., 2005; Wolff et al., 2009). A subgroup of these drug-sensitive mutants, which displayed Multiple Antibiotic Resistance defective phenotypes, was designated as the MARs.
A MAR mutant, MAR2, found in a screen for erythromycin- and vancomycin-sensitive mutants, exhibited additional susceptibility to other classes of antibiotics with diverse chemical structures and functions. These include antibiotics targeting several steps in the cell wall biosynthetic pathways, such as those of the carbapenem subgroup of β-lactam antibiotics, vancomycin and bacitracin; antibiotics targeting protein synthesis, such as those of the macrolide, streptogramin and aminoglycoside classes; antibiotics targeting DNA replication such as novobiocin; as well as those targeting transcription, such as rifampicin which is also a front-line tuberculosis drug (Fig. S2, Table S1). Minimum inhibitory concentrations (MICs) for five representative antibiotics, quantified by using E-test assay (AB Biodisk, Sweden) as previously described (Nguyen et al., 2005; Wolff et al., 2009), indicated that MAR2 was 63-, 1000-, 16-, 80- and 6-fold more susceptible than the wild-type strain to vancomycin, erythromycin, imipenem, rifampicin and amoxicillin/clavulanic acid respectively (Table 1).
Table 1. Susceptibility of M. smegmatis strains to antibiotics.
Genetic mapping using a nested PCR method as previously described (Nguyen et al., 2005) localized the transposon insertion to a gene encoding a putative lipoprotein of unknown function (msmeg_1876) (Fig. 1). The deduced amino acid sequence of the mutated gene had highest similarities to a hypothetical lipoprotein in M. tuberculosis (LpqBTB) and homologues in other mycobacteria. In all available Actinobacterial genome sequences where homologues of LpqB were found, the encoding genes were always clustered with two other genes, mtrA and mtrB, which, respectively, encode for a RR (MtrA) and a HK (MtrB) of a signal transduction system (Fig. S1).
The Himar1 insertion in the lpqB gene of M. smegmatis MAR2 was further confirmed by PCR amplification of the mutant locus, using primers flanking the putative open reading frame. The mutant gene generated a larger fragment corresponding to the inserted transposon (2199 bp). Insertion of the transposon resulted in a decreased mobility of the PCR product on an agarose gel (Fig. 1B). Sequencing of the lpqB-Himar junction region from this PCR product identified the insertion site at the dinucleotide TA517–518 that introduced a stop codon after the triplet encoding the Leu172 residue (Fig. 1A, top). To confirm loss of the lpqB gene product, Western blots were done using a polyclonal antibody raised against a synthetic peptide sequence of LpqB (see Experimental procedures). The antibody recognized a protein band of ∼61 kDa corresponding to the predicted molecular weight of LpqB in the cell lysate of wild-type M. smegmatis, which is absent in the lysate of MAR2, confirming the absence of LpqB expression in MAR2 (Fig. 1C).
Cross-species expression of M. tuberculosis LpqB restores wild-type multidrug resistance to MAR2
To confirm that the multidrug sensitivity phenotype of MAR2 was due to a lack of LpqB activity, the lpqB gene of M. tuberculosis (lpqBTB) was cloned and expressed in the MAR2 mutant. Transformation of the pVN751 [PSOD-lpqBTB] plasmid (Table S2) resulted in restoration of LpqB expression in MAR2 (Fig. 1C). Heterologous expression of LpqBTB (76% identity and 86% similarity to M. smegmatis LpqB) increased drug resistance of MAR2 (Table 1), indicating that the increased antibiotic susceptibility of MAR2 was due to the lpqB deletion. The result also suggests that LpqB provides this function not only to M. smegmatis but also to other mycobacterial species. However, the LpqB proteins from the two bacterial species might carry some levels of specificity because complementation was not complete for some antibiotics (Table 1).
Plasmid pVN751 produced an increased level of LpqB (Fig. 1C). However, this increased expression did not lead to an increased drug resistance (Table 1), suggesting that LpqB does not directly control dedicated drug resistance mechanisms; but rather that its deletion affects principal cellular processes (e.g. cell wall biosynthesis or homeostasis) that influence drug susceptibility. In fact, the overexpression of LpqB from pVN751 per se might have also caused the incomplete complementation of MAR2 due to its reverse effects on downstream gene expression leading to disorders in such cellular processes.
Disruption of LpqB results in altered colonial morphology and cell aggregation
The MAR2 mutant displayed abnormal morphology at several levels. In liquid media, MAR2 culture exhibited an increased cell–cell aggregation. After 2 days of growth in 7H9, the wild-type culture was still well suspended in the medium whereas the MAR2 culture was readily settled at the bottom of culture tubes (Fig. 2A). This reflected an alteration in cell wall composition, surface properties or cellular morphology of MAR2 cells. Altered cell–cell interactions or cell wall surface properties usually give rise to changes in colonial morphology. On solid media, colonies of MAR2 showed a caved-in structure without dense cores (Fig. 2B). This altered colonial morphology was most obvious on 7H10 medium supplemented with Tween-80. Although they were not exactly like wild-type, colonies of the MAR2 strain expressing M. tuberculosis LpqB lost the caved-in structure, characteristic of MAR2 morphology (Fig. 2B, right panel).
Morphological changes associated with increased multidrug sensitivity of the M. avium mtrB mutant correlates with increased Congo Red staining (Cangelosi et al., 2006). In addition, M. avium clinical isolates with an increased Congo Red uptake (red morphotype) also display reduced multidrug resistance and virulence compared with isogenic ‘white’ counterparts (Cangelosi et al., 2001). After 4 days of growth on a solid medium containing Congo Red at 37°C (without Tween-80), a clear phenotypic difference among the M. smegmatis strains was observed, which correlated with LpqB expression. In contrast to the pink and dry colonies of wild-type and the complemented strain, MAR2 colonies were red with shiny wet cores (Fig. 2C), indicating an increased Congo Red uptake. In support of a defective cell wall homeostasis, MAR2 also exhibited an increased sensitivity to the detergent sodium dodecyl sulphate (SDS) when grown on a solid medium. MAR2 was at least 100-fold more susceptible to SDS than wild-type (Fig. 2D). All wild-type morphological characteristics were restored when MAR was complemented with plasmid pVN751 expressing M. tuberculosis LpqB.
Disruption of LpqB affects surface motility and biofilm formation
Because the MAR2 mutant showed an abnormal colonial morphology and cell–cell aggregation, we determined whether LpqB is also required for sliding motility and biofilm formation. As expected, MAR2 was defective in sliding on the surface of minimum medium agar plates, and this defect could be complemented by expression of lpqBTB (Fig. 3A). Studies of M. smegmatis biofilms have established a relationship between surface motility and the ability to form biofilms (Recht and Kolter, 2001; Arora et al., 2008). Wild-type M. smegmatis, MAR2 and the complemented strain were subjected to an air-liquid biofilm assay (Ojha et al., 2005). At day 3 after inoculation, the culture of wild-type M. smegmatis had already colonized the whole surface of the liquid medium, whereas the MAR2 culture only formed unconnected clumps in the liquid phase and a few islands of growth on the surface. At day 5, typical mature biofilm structures appeared on the surface of the wild-type culture, but the formation of those structures was clearly delayed on the surface of MAR2 (Fig. 3B and C).
Disruption of LpqB affects cell division
Microscopic analyses revealed a striking phenotype of the MAR2 mutant. Instead of the normal rod shape of wild-type M. smegmatis cells, MAR2 cells grew as branched filaments (Fig. 4A, white arrows). The length of MAR2 cells was significantly greater than that of wild-type with many cells reaching 20 µm (Fig. 4A and B; Fig. S3). Cross-species expression of lpqBTB restored normal cell length to MAR2 (Fig. 4A, right panel). For quantitative analysis, the length of 100 random cells of wild-type and MAR2 was measured after 24 and 51 h of growth in a liquid medium (Fig. 4B). Whereas wild-type cells had an average cell length limited to a 3–5 µm range at both time points, MAR2 cells exhibited a wide range of lengths, from 3 to 20 µm (Fig. 4B). After 51 h, 75% of MAR2 cells were greater than 8 µm in length. These observations suggested that MAR2 has a defect in cell length or division control.
Cell division control is commonly coupled to DNA replication and segregation. To investigate nucleoid localization in the filamentous MAR2 cells, they were stained with SYTO 11, a membrane permeable dye that fluoresces when bound to nucleic acids. Whereas wild-type M. smegmatis displayed mostly 1–2 fluorescent segments per cell, MAR2 filaments fluoresced in a beaded pattern along the cell length, suggesting that they carry multiple chromosomes (Fig. 4A). Most filaments of MAR2 cells contained from 2 up to 12 fluorescent beads when examined at 24 and 51 h post inoculation (Fig. 4C). It is noteworthy that some of the filaments contained empty sections (Fig. 4A, empty arrows), similar to the effect caused by depletion of yycF in B. subtilis (Fabret and Hoch, 1998). Strikingly, 42% and 50% of the MAR2 filaments examined at these time points had branches respectively. These branched compartments, commonly budded out at loci of putative septa, also contained nucleoids (Fig. 4A, white arrows). This filamentous, branched, polyploid morphology closely resembled Streptomyces, another genus belonging to the taxon Actinobacteria.
lpqB encodes a membrane-bound lipoprotein
Protein sequences of LpqB orthologues revealed the presence of a signal peptide typically present at the N-terminus of lipoproteins, which is required for secretion through the general Sec secretion system (Fig. 5A). To determine the subcellular localization of LpqB, cell fractionation was carried out, followed by Western blot analyses with LpqB antibody. In both M. smegmatis and the pathogenic M. tuberculosis (Fig. 5B, left and right panels respectively), LpqB was predominantly present in the cell membrane and cell wall fractions, suggesting that LpqB function is associated with its localization in membrane or cell wall. This was further supported by the fact that expression of a cytoplasmic form of LpqB failed to restore wild-type phenotypes to the MAR2 mutant (not shown). Next, to validate the lipoprotein nature of LpqB, M. smegmatis strains expressing LpqB were grown in the presence of globomycin, a specific inhibitor of the signal peptidase LspA (Inukai et al., 1978; 1979) required for the removal of signal peptides from lipoproteins after their translocation to the extracytoplamic milieu. A mobility shift of lipoproteins on SDS-PAGE gels usually indicates the inhibitory effect of globomycin on the removal of signal peptides (Gibbons et al., 2007). When wild-type M. smegmatis or the MAR2 strain overproducing LpqBTB was grown in the absence of globomycin and the protein extracts were separated in the conditions described in Experimental procedures, LpqB was detected as two bands migrating closely to each other [Fig. 5C (−)]. However, when the bacterial cultures were treated with globomycin, a clear shift in mobility was observed for the lower band [Fig. 5C (+)], suggesting that the signal peptide processing of LpqB was inhibited. Collectively, these results suggest that LpqB is a lipoprotein that is present in the cytoplasmic membrane and cell wall of the mycobacterial cells.
LpqB interacts with the extracellular domain of MtrB
Phenotypic similarities of MAR2 and mutants of the adjacent genes mtrA and mtrB (Figs 2–4; Möker et al., 2004; Brocker and Bott, 2006; Cangelosi et al., 2006; Fol et al., 2006) suggested that these genes might work together in the same pathway(s). The presence of LpqB in the cytoplasmic membranes and cell wall of M. smegmatis (Fig. 5) further suggested that LpqB activities were more directly related to the membrane-bound MtrB, than the cytoplasmic MtrA. To investigate if LpqB interacts with MtrB, their in vivo association was investigated using a bacterial two-hybrid system (Bacteriomatch, Stratagene). Different domains of MtrB (Fig. 6A, and the far left panel in Fig. 6B) were cloned and expressed as fusions to the C-terminal end of the full-length λcI protein in the bait vector pBT. A cytoplasmic peptide of LpqB (LpqB22–583, without the signal peptide) was fused to the N-terminal domain of the α-subunit of RNA polymerase in the target vector pTRG. Bacteriomatch I Reporter strain (Stratagene) was co-transformed with these fusion constructs. An interaction between bait and target peptides would recruit RNA polymerase to the promoter and activate the transcription of the reporter genes ampR and lacZ that allow growth in the presence of carbenicillin and the production of β-galactosidase respectively. Out of all the combinations of the plasmids (Fig. 6B, first and second panels from left), only combination 2 (positive control, Gal11p + LGF2) and combination 6 (MtrB140–214 + LpqB22–583) produced colonies of the reporter strain grown on carbenicillin plates (250 µg ml−1, CarbR, third panel). This result was further confirmed by the increased production of β-galactosidase by the reporter strain coexpressing the plasmid combinations (Fig. 6B, fourth and far right panel). The MtrB61–214 peptide displayed a weak interaction with LpqB22–583 (Fig. 6B, combination 7, right panel). Compared with the MtrB140–214 peptide that contains no predicted transmembrane domain, MtrB61–214 may carry one (Fig. 6A, model I, residues 120–138) that may localize MtrB61–214 to the membrane proximity, therefore limiting its association with the cytoplasmic LpqB peptide. It is also possible that MtrB61–214 not folding correctly accounts for its weak interaction with LpqB. The facts that the N-terminus of MtrB is highly positively charged (pI 13) and that the second transmembrane helix in model I has a low probability score (Fig. 6A, legend) support the topology in model II. Although the precise membrane topology of MtrB remains to be established experimentally, these results strongly suggest the presence of a sensor domain located within the extracytoplasmic sequence of MtrB that interacts with LpqB.
LpqB affects phosphorylation but not expression of MtrA
To investigate whether LpqB affects expression of MtrA, two different approaches were taken. First, plasmid pVN781 expressing β-galactosidase from the mtrA promoter was transformed to wild-type M. smegmatis and MAR2 by electroporation. Transformants were grown on 7H10-hygromycin medium and samples were collected during growth to determine β-galactosidase activity (see Supporting information). There were no differences in mtrA promoter activity observed between wild-type and MAR2 (Fig. 7A, upper panel). Second, plasmid pVN762 expressing c-Myc tagged MtrA from its native promoter (Table S2) was introduced to wild-type and MAR2 and the expression was analysed by Western blot using a c-Myc antibody. Cultures of M. smegmatis strains were harvested during growth and equal amounts of cellular proteins from each sample were separated on SDS-PAGE, followed by Western blot (Fig. 7B). Furthermore, to investigate whether LpqB mediates MtrA expression to respond to cell wall or general antibiotic stress, mycobacterial cultures were exposed to vancomycin (10 µg ml−1) or erythromycin (100 µg ml−1) for 1 h before samples were prepared (not shown). Expression of MtrA from its native promoter was not affected by LpqB in any of the conditions investigated. These results indicate that LpqB has no effect on MtrA at the expression level.
To investigate whether LpqB affects the phosphorylation status of MtrA, M. smegmatis strains expressing c-Myc tagged MtrA from pVN762 were labelled in vivo by [32P]-orthophosphate, followed by immuno-purification of c-Myc tagged peptides. Purified materials were separated by SDS-PAGE, blotted onto Whatman filter paper and examined by autoradiography. To retain the stability of the phospho-aspartate, samples had only been treated by mild heating (see Experimental procedures) before SDS-PAGE. Under this condition, the c-Myc tagged MtrA migrated as two forms, 25 and 50 kDa (Fig. 7C). The higher molecular weight form of MtrA (possibly MtrA homodimers) was disrupted when samples were boiled (Fig. S4). More importantly, this form of MtrA was [32P]-labelled, suggesting that it was phosphorylated in vivo. By contrast, there was no phosphorylation signal detected for the monomer form of MtrA (Fig. 7C). The phosphorylated form of MtrA (MtrA-P) accounted for 38.17% of the total MtrA signal in wild-type M. smegmatis but only 4.93% in MAR2 (Fig. 7C). This result indicates that LpqB regulates phosphorylation levels of MtrA.
Phosphorylation status of MtrA mediates LpqB function
MtrA belongs to the OmpR/PhoB subfamily of RRs that are characterized by a winged helix–turn–helix DNA binding domain. Phosphorylation of the regulatory domain of these regulators often correlates with induced dimerization, which enhances their ability to bind DNA and regulate transcription (Igo et al., 1989; Fiedler and Weiss, 1995). A crystal structure revealed that MtrA has the typical C-terminal effector DNA binding domain as well as an N-terminal signal receiver domain with the conserved phospho-acceptor site D56 (Friedland et al., 2007). This aspartate residue is highly conserved and is in close proximity to the active site of MtrA (Friedland et al., 2007). It has been established that the phosphotransfer from a HK induces conformational changes that modulate the activities of its paired RRs. To assess the in vivo functions of D56 in MtrA, the mtrA gene and its D56A or D56E mutant alleles were cloned and overexpressed in wild-type M. smegmatis and MAR2. Similar experiments with other RRs predicted that MtrA(D56A) would not be activated by phosphotransfer (Parkinson and Kofoid, 1992) and thus will remain in an inactive form. In contrast, because the structure of glutamate mimics the phosphorylated aspartate residue, the MtrA(D56E) mutant might have a more active conformation, independent of the upstream signal transduction (Parkinson and Kofoid, 1992). To further avoid possible feedback regulation or interferences by third-party systems, these mutated MtrA alleles were in trans expressed from the strong PSOD promoter (pVN747) instead of the native promoter. Whereas in trans expression of the MtrA(D56A) allele (pVN765) in MAR2 did not lead to any detectable changes in its abnormal phenotypes, the expression in wild-type M. smegmatis resulted in striking phenotypes resembling many of those described above for MAR2. The M. smegmatis/pVN765 cells also displayed an increased susceptibility to antibiotics, increased cell–cell aggregation (Fig. S5), as well as significant defects in cell division control and biofilm growth (Fig. 7D). In contrast, when plasmid pVN766 expressing the MtrA(D56E) allele from the same promoter was transformed to both wild-type and MAR2, opposite effects were observed. MAR2 transformants expressing MtrA(D56E) displayed wild-type characteristics whereas wild-type M. smegmatis remained unchanged (Fig. 7D). Expression of wild-type MtrA (pVN763) in MAR2 led to effects that are similar to, but less intense than, those caused by the expression of MtrA(D56E) (Fig. S6). Together, these results indicate that the phosphorylation status of MtrA mediates functions of LpqB.
LpqB affects promoter activity of dnaA– an mtrA regulon gene
The only direct target of MtrA in mycobacteria thus far known is the gene encoding the master regulator of DNA replication DnaA (Fol et al., 2006). To further investigate whether lpqB expression affects MtrAB function, we measured the promoter activity of dnaA in the M. smegmatis strains during growth on solid medium 7H10. During earlier stages of growth (48 and 72 h after inoculation), dnaA promoter activity of MAR2 was elevated compared with that of wild-type M. smegmatis (Fig. 7A, lower panel). These results indicate that LpqB negatively affects dnaA expression through a control of its promoter activity.
Computational analyses identified approximately 48 and 61 putative lipoproteins encoded in the genomes of M. tuberculosis and M. smegmatis respectively (Rezwan et al., 2007). These proteins provide diverse functions in mycobacterial species, including solute binding proteins for efflux pumps, carriers for lipid translocation, enzymes involved in biosynthesis, degradation and metabolism of cell wall constituents, as well as modulators of host cell adhesion and pathogenesis of pathogenic mycobacteria (Sutcliffe and Harrington, 2004). There is growing evidence that lipoproteins may also function in signal transduction and regulatory processes in Actinobacteria, including Mycobacterium (Steyn et al., 2003). In S. coelicolor, the extracytoplasmic lipoprotein CseA was suggested to modulate activity of the two-component system CseB-CseC through interactions with the extracellular sensor domain of the HK CseC to respond to cell wall stress (Hong et al., 2002; Hutchings et al., 2006). Analysis of Actinobacterial genomes revealed a significant number of HKs that are genetically linked with a putative lipoprotein, suggesting that such ‘three-component’ signal transduction systems (i.e. two components plus a lipoprotein) are common in high G+C Gram-positive bacteria (Hoskisson and Hutchings, 2006). The putative tricistronic operon encoding MtrA–MtrB–LpqB, a member of this proposed three-component system family, is ubiquitously present in Actinobacteria (Fig. S1) but absent in other bacterial groups, suggesting that it provides specialized functions to these bacteria (Gao et al., 2006). Although the function of MtrAB has been studied at some level, nothing is known about the lipoprotein LpqB. Here, we present evidence supporting that LpqB functions as an accessory protein that modulates activity of the MtrAB system in controlling homeostasis of the cell wall and cell division (Fig. S7).
In Corynebacterium glutamicum, both mtrA and mtrB could be deleted whereas deletion of lpqB was lethal (Brocker and Bott, 2006), suggesting that the gene is essential in this bacterium. Likewise, a transposon site hybridization experiment suggested that lpqB is also indispensable in M. tuberculosis (Sassetti et al., 2003). We show here that an lpqB transposon mutant could be isolated from the fast-growing M. smegmatis. A transposon insertion in lpqB abolished translation of almost two-thirds of the protein from the C-terminus (Fig. 1). Although surviving the disruption, the lpqB mutant (MAR2) exhibited a pleiotropic phenotype that relates to defects of the cell wall and cell division. Similar to the mtrB mutant of M. avium (Cangelosi et al., 2006), MAR2 displayed a greatly enhanced susceptibility to multiple antibiotics with a diverse array of chemical structures and mechanisms of action (Fig. S2, Table S1). This non-specific antibiotic sensitivity spectrum suggests that the cell wall integrity of MAR2 is compromised.
Interrupted cell wall homeostasis often leads to collateral phenotypes such as alterations in surface motility and biofilm formation (Senadheera et al., 2007). MAR2 displayed an increased cell–cell aggregation in liquid media (Fig. 2A), which is likely the cause of the respective sliding motility (Fig. 3A) and biofilm growth (Fig. 3B and C) defects. A recent study in M. tuberculosis has revealed a correlation of biofilm growth and the development of persisters with increased antibiotic tolerance (Ojha et al., 2008). This correlation may underlie the increased drug tolerance observed in latent or relapsed tuberculosis. Given its role in a signal transduction pathway that is required for both antibiotic resistance and biofilm growth of Mycobacterium, LpqB and its associated signal transduction network might function as a phenotypic switch contributing to the development of drug-tolerant persisters during biofilm growth and latent infection of M. tuberculosis.
Cytokinetic and cell wall homeostatic controls might also be co-regulated by common genetic determinants. For example, the multiple-component signal transduction system YycFGHIJ in low G+C Gram-positive bacteria co-ordinates cell division with cell wall biosynthetic processes (Bisicchia et al., 2007; Fukushima et al., 2008; Winkler and Hoch, 2008). Being a component of the septal cell division protein complex allows the HK YycG to adjust its phosphotransfer to the RR YycF in a spatial manner, thereby affecting gene expression involved in cell wall synthesis in co-ordination with cell division processes (Fukushima et al., 2008). In support of this paradigm, MAR2 exhibited defects not only in cell wall integrity but also cell division control. Significant increase of the average cell length, together with great length variation, suggests that either spatial or temporal control of cell division was defective (Fig. 4). Again, this phenotype is similar to the phenotypes of the mtrAB deletions. Disruption of either mtrA (Möker et al., 2004) or mtrB (Brocker and Bott, 2006; Cangelosi et al., 2006) leads to an increased cell length in C. glutamicum and M. avium. Interestingly, deletion of both mtrA and mtrB in C. glutamicum increases cell length to a greater extent than either the single deletion (Brocker and Bott, 2006). Together, these results indicate that both MtrAB and LpqB are required for the control of cell length in Mycobacterium, as well as other rod-shaped Actinobacteria. Thus far, it is unclear how cell length is determined in this group of bacteria, which apparently lack the cell division positioning proteins MinCD, as well as the cell shape determining MreB-like proteins (Jones et al., 2001). Future study will need to identify the nature of the upstream factors that provide the ‘cell length’ signals to the LpqB-MtrAB system, as well as the downstream regulons responsible for cell division.
Another intriguing feature of MAR2 is its morphological reminiscence of Streptomycetes (Fig. 4). Altered expression of certain cell division proteins had previously been shown to switch Mycobacterium from the typical bipolar growth to multi-polar growth and branch formation (Gomez and Bishai, 2000; Nguyen et al., 2007; Scherr and Nguyen, 2009). This morphological switch possibly underlies the observed pleomorphism of many Mycobacterium species that might help them to adapt to environmental changes, including the niche within infected host cells (Chauhan et al., 2006; Scherr and Nguyen, 2009).
Whereas in vitro study of Corynebacterium suggested that detection of osmotic stress is mediated exclusively by the cytoplasmic domain of MtrB (Möker et al., 2007), our results showed that LpqB only interacts with the extracytoplasmic domain (Fig. 6), indicating that the latter domain might play an important role in receiving signals from LpqB. The discrepancy is not surprising because functional variations of MtrAB have been observed in different species of Actinobacteria. For example, our experiments did not reveal a role of LpqB in culture pH control as shown for the mtrAB mutants of Corynebacterium (Fig. S8) (Möker et al., 2004). Providing cells with such a diverse array of functions would require MtrB to evolve multiple mechanisms both for the signal reception from various upstream sources, as well as for the signal delivery to multiple downstream partners.
The interaction of LpqB and the sensor domain of MtrB probably triggers autophosphorylation of the cytoplasmic domain of MtrB, leading to the activation of MtrA and downstream signalling cascades (Fig. S7). We show that whereas LpqB deletion did not affect MtrA expression (Fig. 7A and B), it altered the phosphorylation status of the protein (Fig. 7C). This phosphotransfer deficit was likely rooted in a failed activation of MtrB in response to unknown physiological or environmental conditions. In support of this model, targeted mutagenesis experiments indicated that the phosphorylation status of MtrA mediates the LpqB-related characteristics (Fig. 7D and Fig. S5). Overexpression of MtrA(D56A) from PSOD in wild-type M. smegmatis led to MAR2-like phenotypes (Fig. 7D) that are similar to the phenotypes described for the mtrAB mutant of C. glutamicum (Möker et al., 2004). These results suggest that the D56A form of MtrA was able to create a dominant negative effect when in trans overexpressed in wild-type mycobacteria. It remains to be established how the increased presence of the inactive form of MtrA leads to this dominant negative effect. Given the low probability of inactive MtrA binding DNA targets or forming higher order complexes, as revealed by its crystal structure (Friedland et al., 2007), we favour the hypothesis that MtrA(D56A) interacts with the intracellular HK transmitter domain of MtrB, thereby outcompeting the wild-type MtrA in the phosphotransfer reaction. Similar mechanisms of signal transduction delays have been reported previously with other two-component systems (Goymer et al., 2006). LpqB-mediated activation of MtrAB is apparently required for a suppression of dnaA expression as disruption of lpqB led to increased dnaA promoter activities (Fig. 7A). In fact, MtrA is able to function as either an activator or a repressor of gene transcription (Brocker and Bott, 2006). In M. tuberculosis cells growing in macrophages, overexpression of wild-type MtrA resulted in dramatic upregulation of dnaA expression (Fol et al., 2006). This led the authors to conclude that phosphorylated MtrA is an activator of dnaA transcription (Fol et al., 2006). Contradictory to that conclusion, simultaneous overexpression of MtrB relieves the dnaA upregulation caused by the overexpression of MtrA alone (Fol et al., 2006), suggesting that the MtrB-mediated phosphorylation of MtrA is required to suppress dnaA expression. The crystal structure of MtrA reveals an extensive interface between the N-terminal phospho-receiver and the C-terminal DNA binding domain which stabilizes the inactive conformation, and thereby may restrict the phosphorylation rate of MtrA (Friedland et al., 2007). In certain conditions when the upstream phosphotransfer rate is limited, an unmatched increased production of wild-type MtrA (Fol et al., 2006) may lead to a reduced ratio of active/inactive forms of MtrA [similar to the overexpression of MtrA(D56A)], thus lowering the MtrA-mediated suppression of dnaA transcription. Together with previous work (Fol et al., 2006; Friedland et al., 2007), our results suggest that LpqB is an activator of the MtrAB system that negatively controls dnaA expression. Its analogous function and mechanism of action (Fig. S7) further support the hypothesis that MtrA–MtrB–LpqB is the Actinobacterial analogue of the YycFGHIJ system in Firmicutes (Winkler and Hoch, 2008).
The absence of two-component signal transduction systems in animals has generated interest in their potential as targets for future chemotherapies attacking drug-resistant infections (Okada et al., 2007; Winkler and Hoch, 2008). In view of the essentiality of MtrAB in mycobacterial viability, antibiotic resistance and biofilm growth, this system and its accessory proteins such as LpqB may serve as promising targets for future development of new antibiotics that ameliorate the current shortage of effective chemotherapies against multidrug-resistant and extensively drug-resistant tuberculosis (Nguyen and Pieters, 2009).
Bacterial strains, chemicals, media and growth conditions
All strains and plasmids used in this study and the details of their constructions can be found in the Supporting information. Wild-type M. smegmatis and its transposon-derived mutants were grown in 7H9 liquid medium or on 7H10 (Difco) or Luria-Bertani (LB) agar medium supplemented with 0.5% Tween 80. Kanamycin was used at a final concentration of 50 µg ml−1. Hygromycin was used at 100 and 75 µg ml−1 for Escherichia coli and mycobacteria respectively. Globomycin was a generous gift from Masatoshi Inukai (Sankyo Corporation and International University of Health and Welfare, Japan). Preparation of competent cells and transformation were carried out as described (Braunstein et al., 2002). The c-Myc monoclonal antibody (9E10) was obtained from the Developmental Studies Hybridoma Bank, University of Iowa.
Site-directed mutagenesis of D56 in MtrA
The Expand Long Template PCR kit (Roche Molecular Biochemicals) was used to amplify the M. smegmatis mtrA alleles from pVN762. Mutant alleles of mtrA in which the phosphorylated residue aspartate 56 (D56) was replaced by an alanine (D56A) or a glutamate residue (D56E) were generated by a two-stage PCR procedure. Primers MtrA2.N and MtrA4.H were used together with either the primer pairs (mtrA-DA1 + mtrA-DA2) or (mtrA-DE1 + mtrA-DE2) to amplify the D56A or D56E allele respectively (Table S3). All PCR products were cloned in the vector pGEM-T Easy and sequences were confirmed by sequencing. Mutant mtrA genes were cloned into pVN747 (NdeI/HindIII), thus their translation was fused to the PSOD promoter upstream (pVN765 and pVN766, Table S2).
Inhibition of signal peptidase-mediated processing of LpqB
Mycobacterium smegmatis cultures (OD600∼2) were split into two portions (20 ml) and grown for 16 h in the presence or absence of 50 µg ml−1 globomycin. Cell lysates were prepared by bead beating using the Fastprep 24 (MP Biomedicals, Solon, OH, USA) and separated by SDS-PAGE. To facilitate separation of LpqB forms, samples were electrophoresed for an extended time until proteins of molecular weights smaller than 30 kDa had run out of the gels. Isoforms of LpqB were visualized by Western blot using the anti-LpqB antibody.
Subcellular fractions of M. smegmatis cells were prepared as described (Gibbons et al., 2007). Briefly, mycobacterial cultures (500 ml) were collected and washed with PBS (3000 g, 10 min). Cell pellets were resuspended in 10 ml of PBS supplemented with protease inhibitors and disintegrated by French press. Unbroken cells were removed and the total cell lysate was centrifuged at 27 000 g for 30 min to pellet the cell wall. The supernatant was centrifuged at 100 000 g for 2 h to separate the membrane fraction from the soluble fraction.
Bacterial two-hybrid protein interaction assay
Protein interaction assays were performed using the BacterioMatch I reporter strain, following the manufacturer's instructions (Stratagene). Plasmids expressing prey and bait peptides were constructed as described in the Supporting information. The strength of interaction was quantified by analysing the β-galactosidase activity of the reporter strains and expressed as Miller Units.
In vivo phosphorylation and immuno-precipitation
Plasmid pVN762 expressing c-Myc tagged MtrA from its native promoter was transformed to wild-type M. smegmatis and MAR2. Expression of c-Myc tagged MtrA was detected by Western blot, using anti-c-Myc antibody. In vivo[32P] labelling was performed as previously described (Radhakrishnan et al., 2008) with modifications. A single colony of cells picked from a 7H10 plate was washed and grown overnight in 7H9 medium to an OD600 of 0.5. This culture was used to inoculate 50 ml 7H9 medium and grown to an OD600 of 0.8, washed and resuspended in 1 ml phosphate-depleted Sauton's medium. The culture was then labelled for 1 h at 37°C using 100 µCi of [γ-32P]-orthophosphoric acid (Perkin Elmer). Following cell lysis by bead beating, proteins were immuno-precipitated with the ProFound c-Myc Tag IP/CoIP Kit according to the manufacturer's instructions (Pierce, Rockford, IL, USA). The bound materials were eluted from the columns followed by a mild heating (37°C 15 min followed by 60°C 2 min). Samples were resolved by SDS-PAGE, gel dried onto Whatman Paper, and [32P]-labelled MtrA was quantified using a Storm 820 PhosphorImager and ImageQuant software version 4.0 (Molecular Dynamics) and normalized to the relative MtrA content as determined by immunoblotting of the same immuno-precipitated materials.
Anti-LpqB polyclonal antibody was produced in rabbits by using a synthetic 15-amino-acid peptide sequence identical in M. smegmatis and M. tuberculosis LpqB proteins (MDPDVLLREFLKATA) (Affinity BioReagents, Golden, CO, USA). The antibody was purified by affinity chromatography using the synthetic peptide as a ligand.
Details of other procedures can be found in the Supporting information.
We thank Michael Niederweis, Masatoshi Inukai, Elliott Crooke, David Alland, Sabine Ehrt and Dirk Schnappinger for providing materials, Bing Liu, Daniel Kiss and Megan Mamolen for technical assistance, Piet de Boer, Kien Nguyen and Abram Stavitsky for critical reading of the manuscript. This work was supported by start-up funds from the School of Medicine, a STERIS Infectious Diseases Research Award and CFAR Developmental Awards from the Case/UH Center for AIDS Research (AI36219) to L.N. S.O is a trainee of the Fogarty AIDS International Training and Research Program (AITRP) at Case School of Medicine.