Mycobacteria are among the clinically most important pathogens, but still not much is known about the mechanisms of their cell cycle control. Previous studies suggested that the genes encoding ParA and ParB (ATPase and DNA binding protein, respectively, required for active chromosome segregation) may be essential in Mycobacterium tuberculosis. Further research has demonstrated that a Mycobacterium smegmatisparB deletion mutant was viable but exhibited a chromosome segregation defect. Here, we address the question if ParA is required for the growth of M. smegmatis, and which cell cycle processes it affects. Our data show that parA may be deleted, but its deletion leads to growth inhibition and severe disturbances of chromosome segregation and septum positioning. Similar defects are also caused by ParA overproduction. EGFP–ParA localizes as pole-associated complexes connected with a patch of fluorescence accompanying two ParB complexes. Observed aberrations in the number and positioning of ParB complexes in the parA deletion mutant indicate that ParA is required for the proper localization of the ParB complexes. Furthermore, it is shown that ParA colocalizes and interacts with the polar growth determinant Wag31 (DivIVA homologue). Our results demonstrate that mycobacterial ParA mediates chromosome segregation and co-ordinates it with cell division and elongation.
Due to their enormous human health impact, bacteria belonging to the genus Mycobacterium, particularly Mycobacterium tuberculosis, have mainly been studied in the contexts of pathogenicity and infection. Critical for the treatment of M. tuberculosis infection is the ability of these bacteria to exist in a dormant, non-replicating state in which they remain metabolically active and are able to return to their normal cell cycle. Although this state is closely connected with persistence of M. tuberculosis and lowered susceptibility of these bacteria to drugs, particularly those targeting cell division and cell wall synthesis, the mechanisms controlling the cell cycle during dormancy have not yet been fully explored (Wayne and Sohaskey, 2001). Until now, studies on mycobacterial cell cycle processes have largely focused on the main proteins involved in chromosome replication, cell division and cell elongation, i.e. DnaA (Yamamoto et al., 2002; Zawilak et al., 2004; Zawilak-Pawlik et al., 2005; Madiraju et al., 2006; Kumar et al., 2009), FtsZ (Dziadek et al., 2002, 2003; Rajagopalan et al., 2005a,b; Dziedzic et al., 2010) and the DivIVA homologue, Wag31 (Nguyen et al., 2007; Kang et al., 2008) respectively. Unlike in Bacillus subtilis, where DivIVA interacts with MinC and MinD to control septum formation, it is mainly responsible for controlling cell wall synthesis at the cell poles in Mycobacterium (Nguyen et al., 2007; Kang et al., 2008). Sequence analysis of the M. tuberculosis (Hett and Rubin, 2008) and Mycobacterium smegmatis genomes revealed a lack of genes encoding some cell division regulators, such as FtsA and the Min system proteins. Studies of the co-ordination of key steps of the Mycobacterium cell cycle (i.e. chromosome replication and segregation, cell division and elongation) should shed light on persistence of tuberculosis infection.
Up to now the question of Mycobacterium chromosome segregation has been addressed by the studies of segregation proteins ParB (Jakimowicz et al., 2007a; Chaudhuri and Dean, 2011) and ParA (Nisa et al., 2010), which are components of the active chromosome segregation machinery found in a number of bacterial species (e.g. B. subtilis, Caulobacter crescentus, Vibrio cholerae, Pseudomonas spp., Corynebacterium glutamicum, and Streptomyces coelicolor). In all studied bacteria, ParB homologues bind to a cluster of DNA sequences called the parS sites, which range in number from 3 in V. cholerae and C. glutamicum to about 20 in S. coelicolor and are usually localized near oriC (Bartosik and Jagura-Burdzy, 2005). Upon binding to parS sites, ParB oligomerizes to form large nucleoprotein complexes called segrosomes. Segrosomes organize the oriC proximal region of newly replicated chromosomes and facilitate their proper positioning in the future daughter cells, either at the cell centre (in B. subtilis) or close to the pole(s) (in C. crescentus, C. glutamicum and V. cholerae chromosome I) (Leonard et al., 2005; Chaudhuri and Dean, 2011; Chaudhuri et al., 2011). This process is followed by segregation and condensation of other regions of the chromosome (Glaser et al., 1997; Fogel and Waldor, 2006; Donovan et al., 2010; Shebelut et al., 2010).
The formation and positioning of ParB complexes depends on ParA homologues. These are Walker ATPases, whose activity is enhanced in the presence of ParB. ParAs are able to dimerize and polymerize into filaments in a process that is ATP binding-dependent and requires the presence of DNA in some cases (e.g. Thermus thermophilus) but not others (e.g. S. coelicolor) (Leonard and Grimwade, 2005; Ditkowski et al., 2010). Most ParA homologues, such as those in C. crescentus (Ptacin et al., 2010), V. cholerae (Fogel and Waldor, 2006) and C. glutamicum (Donovan et al., 2010), show dynamic localization and extend from one pole towards the middle of the cell; however, the B. subtilis Soj protein (ParA homologue) seems to be associated with septa and with the chromosomal origin regions (Murray and Errington, 2008). ParA filaments undergo ParB-dependent polymerization/depolymerization, and are believed to provide motive force for segrosomes within bacterial cells. A growing body of evidence suggests that ParA proteins link chromosome segregation with other cellular processes, including the initiation of chromosome replication in B. subtilis (Murray and Errington, 2008; Scholefield et al., 2012) and V. cholerae (Kadoya et al., 2011), cell division in C. crescentus and S. coelicolor (Mohl et al., 2001; Jakimowicz et al., 2007b), polar growth in S. coelicolor (B. Ditkowski, submitted) and motility in Pseudomonas aeruginosa (Lasocki et al., 2007).
Even though the parA and parB genes were shown to be essential only in C. crescentus, in most studied bacteria elimination of the ParB protein results in the formation of anucleate cells after division, at frequencies ranging from 1% in B. subtilis (Ireton et al., 1994) to 43% in a C. glutamicum parB deletion strain grown on minimal medium (Donovan et al., 2010). A transposon mutagenesis study performed by Sassetti et al. (2003) indicated that the parAB genes could not be deleted from M. tuberculosis. In M. smegmatis, which is a relatively fast growing model organism often used for studies on mycobacterial cell biology, parB was found to be non-essential, but its deletion resulted in chromosome segregation defects and generated 10% anucleate cells (Jakimowicz et al., 2007a). In M. smegmatis, ParB was shown to bind three parS sites localized around the oriC region in vivo and in vitro, and this binding was enhanced in the presence of ParA (Jakimowicz et al., 2007a). Lowering the level of ParA was shown to decrease the growth rate of M. smegmatis (Nisa et al., 2010), but the function of ParA in the cell cycle of Mycobacterium has not yet been studied more extensively.
Here, the relative scarcity of knowledge of the mycobacterial cell cycle, particularly in terms of chromosome dynamics and cell division, prompted us to address the role of ParA in chromosome segregation and its co-ordination with other cell cycle processes in M. smegmatis.
In M. smegmatis, parA is non-essential but required for normal growth
The work of Sassetti et al. (2003) suggested that ParA could not be eliminated from M. tuberculosis H37Rv while our previous studies indicated that Mycobacterium ParA protein, as a component of the partitioning system, may play an important role in the assembly of ParB–DNA complexes (Jakimowicz et al., 2007a). This notion and the complex functions of ParA homologues in other bacteria led us to examine the role of ParA during the cell cycle of Mycobacterium. To analyse if parA (gene msmeg 6939) is essential for the viability of M. smegmatis, a two-step recombination protocol (Parish and Stoker, 2000) was used to generate an unmarked deletion of the parA gene in the M. smegmatis chromosome. The deletion was constructed successfully and the obtained ΔparA strain (KG22; Table 1) was verified by PCR, Southern blotting (data not shown) and Western blotting (Fig. S1). RT-PCR transcript analysis (data not shown) and Western blotting confirmed that the level of ParB was only marginally affected in the parA deletion strain (Fig. S1). Although parA was found to be non-essential for the viability of M. smegmatis, growth rate analysis of the ΔparA strain in rich liquid medium 7H9 + OADC (see Experimental procedures) revealed that the lack of ParA substantially inhibited cell growth (Fig. 1A and B). To verify if the observed growth delay was solely dependent on parA deletion, a complementation strain was constructed. Since parA is believed to be transcribed from its own promoter (Casart et al., 2008), the parA gene and its promoter region (332 bp upstream of the start codon) were cloned into vector pMV306. In the complementation strain, KG23 (Table 1), the obtained plasmid pMV306pnatparA was integrated into the phage L5 attachment site of the parA deletion strain (KG22). The growth curve of complementation strain KG23 did not differ significantly from that of the wild-type strain (Fig. 1A), proving that the growth defect observed in KG22 cells resulted solely from deletion of parA.
M. smegmatis mc2 155 ΔparA attBL5:: pMV306pnatparA
M. smegmatis mc2 155 egfp–parA, parB–mcherry
M. smegmatis mc2 155 attBL5:: pMV306pnatdivIVA–mcherry
M. smegmatis mc2 155 ΔparA, parB–mcherry
M. smegmatis mc2 155 ΔparA attBL5:: pMV306pnategfp–parAS
Elimination of ParA affects chromosome segregation and cell length
To study the influence of ParA on chromosome segregation, the parA deletion strain (KG22) was examined microscopically. DNA staining (DAPI) of cells collected in the exponential growth phase showed a larger fraction of anucleate cells for ΔparA (30.5% of 2162 cells analysed) compared to the wild type (0.2% of 2013 cells analysed) (Table 2, Fig. 1D). In the complementation strain (KG23), less than 4% of cells were anucleate (2118 cell analysed; Table 2, Fig. 1D), confirming that the severe segregation defect in KG22 was due to elimination of ParA. The slight missegregation in the complementation strain may have arisen from the insertion of an additional parS site within the parA gene delivered in trans at the phage L5 attachment site, which is far from the oriC region.
Table 2. Frequencies of anucleate cells in various M. smegmatis strains observed under the microscope (DAPI staining)
% of anucleate cells (∼ 2000 cells counted)
M. smegmatis mc2 155
Interestingly, microscopic analysis of the parA deletion strain revealed an increased variation in cell length. There was a substantial fraction of shortened, usually anucleate cells and a large number of elongated cells compared to the wild type [of ∼ 2000 cells analysed for each strain, 33% of ΔparA (KG22) and 16% of wild-type cells were longer than 6 μm; Fig. 1C].
These results indicated that, compared to the elimination of ParB (10% anucleate cells; Jakimowicz et al., 2007a), elimination of ParA has more severe effects on chromosome segregation in M. smegmatis. Moreover, they revealed that ParA elimination also influenced cell length of M. smegmatis.
ParA overproduction has phenotypic effects similar to those of parB deletion
Our earlier studies showed that both elimination and overexpression of parB reduced the growth rate of M. smegmatis, but overexpression did not affect chromosome segregation (Jakimowicz et al., 2007a). This could suggest that proper maintenance of the ratio between ParA and ParB is important for the cell cycle of M. smegmatis. Therefore, we investigated whether the overexpression of parA could affect growth and/or chromosome segregation in M. smegmatis. An additional copy of the parA gene under the control of the inducible pami promoter in the pMV306 vector was integrated into the chromosome of the wild-type strain to yield the KG11 strain (Table 1). The induction (with 0.1% acetamide) of parA expression in KG11 elevated the protein level of ParA to approximately 20-fold that seen in the wild-type strain, as estimated by Western blotting (Fig. S1). parA overexpression led to growth inhibition (Fig. 1A) and chromosome missegregation (9.3% anucleate cells; Table 2), which was similar to the defects observed for the parB deletion strain. Additionally, we observed an increased variation in cell length (Fig. 1C); 46.3% of ParA-overproducing cells were longer than 6 μm compared to only 24% of cells in the control strain (KG08 harbouring insert-free pMV306 grown on acetamide, or the strain expressing mcherry from the pami promoter in pMV306; data not shown). Thus, parA overexpression affects chromosome segregation in a manner similar to ParB elimination, and also influences cell length.
Deletion and overexpression of parA affect cell division
The high variation of cell length in M. smegmatis strains lacking or overexpressing parA could be the result of impaired cell elongation or division (e.g. changes in the positioning and/or timing of septation). Moreover, disturbances in cell division in the parA mutant could account for its growth inhibition and severe chromosome missegregation phenotypes, which exceeded the segregation defects caused by parB deletion. In order to test the influence of ParA on septum formation, parA deletion (KG22) and wild-type cultures were harvested at the same point of exponential growth phase (OD = 0.4) and subjected to cell wall and membrane staining (with Vancomycin-BODIPY and FM4-64 respectively) (Fig. 2A). Septum position measurements showed that the localization of the septum was highly variable in the ΔparA strain, whereas septa were typically localized within 40–60% of the cell length in the wild type (Fig. 2B).
Septum position was also analysed in the parA-overexpressing strain (KG11) and compared to the control strain (containing an insert-free integrative pMV306 plasmid; KG08). Both strains were grown in the presence of 0.1% acetamide, and samples were prepared for microscopy using cultures from the exponential phase of growth (OD = 0.4). Septum positioning was found to be disturbed in KG11 cells compared to KG08 cells (Fig. 2B), albeit not as severely as in the parA deletion strain.
Thus, the modifications of ParA level appear to affect also the positioning of the septum and possibly the timing of its formation which may be the result of defective chromosome organization.
In parA deletion mutants, the segregation phenotype is suppressed by parB deletion, whereas the other defects are not
To examine which of the defects observed in the parA deletion strain are dependent on ParB, a double-deletion strain was constructed. The ΔparAB strain, KG03 (Table 1), was constructed by two-step recombination and verified by PCR, Southern blotting (data not shown) and Western blotting (Fig. S1). The growth of the ΔparAB strain was impaired, although the CFU (colony-forming unit) count showed that this impairment was not as severe as that of the parA deletion strain (Fig. 3A and B). Surprisingly, microscopic analysis showed that the fraction of anucleate cells in the ΔparAB strain was very low (1.2% of 2014 cells counted; Table 2). Thus, the chromosome segregation defect in the ΔparAB strain was diminished in comparison to the parA or parB single-deletion strain (30% and 10% respectively). In the ΔparAB strain aberrations of cell length were also observed (Fig. 3C); 26% of cells were longer than 6 μm. Cell length abnormalities were only slightly reduced in comparison to the parA deletion strain (33% cells were longer than 6 μm) and still noticeable in comparison to the control strain (16%), indicating that the effect of ParA on cell length may be independent of ParB.
Thus, surprisingly, although the parA phenotype exceeded the parB phenotype in terms of segregation defects, parB deletion suppressed segregation defects caused by the elimination of ParA, but did not significantly alter the other defects, such as disturbed cell length.
EGFP–ParA forms pole-associated complexes with a patch of fluorescence in between
To examine subcellular localization of ParA in M. smegmatis cells, strain KG02 was constructed, in which parA was replaced with an egfp–parA fusion at the original locus (Table 1). Western blotting demonstrated that the expression level of the EGFP–ParA fusion protein was similar to that of ParA in the wild-type strain (Fig. S1), but it also showed that in addition to EGFP–ParA the wild-type ParA protein was also present in KG02. The observed expression of a native parA gene may result from start codon misannotation (the alternative start codon for parA gene is located 102 bp downstream of the annotated start codon, and may be the start of translation due to the presence of an intact ribosome binding site in the egfp–parA fusion gene). The fusion protein did not disturb the growth or chromosome segregation of KG02 cells (1.2% anucleate). Additionally, a new strain was constructed (KG56), in which the pMV306 plasmid containing egfp fused to the downstream start codon of parA, expressing EGFP–ParAS, was integrated into the chromosome of the parA deletion strain (Table 1). In the strain KG56 no wild-type ParA could be detected (Fig. S1) and EGFP–ParAS was fully functional as demonstrated by the complementation of the parA deletion phenotype (1.2% anucleate cells). Microscopic analyses performed on live cells collected from liquid 7H9 + OADC cultures and re-suspended in PBS revealed that the localization patterns of EGFP–ParA and EGFP–ParAS were identical. Both proteins were visible as foci at the cell pole(s), with extended patches of diffuse fluorescence (Figs 4). In the majority of the cells (76% of 5461 and 84% of 1020 cells analysed in strains KG02 and KG56 respectively), intensive fluorescence was visible at both poles. The other cells contained either only a single complex at one pole or three foci. The additional ParA complex in the middle was mostly visible in cells longer than 6 μm (average cell length of 5 μm) (Fig. 4A and E). Measurements of the fluorescence intensity of EGFP–ParA indicated that cells with bipolar complexes frequently had unequal fluorescence intensities at the two poles; only 27% of cells (1218 KG02 cells analysed; Fig. 4C) showed identical intensities at both poles. To address the question if the localization of ParA complexes is affected by the elimination of ParB, strain KG13 was constructed to express egfp–parA in the parB deletion background (Table 1). The level of EGFP–ParA in KG13 was similar to that in KG02, and the distribution of ParA complexes was not visibly affected (data not shown). However, still images analysis does not fully address the question of ParA dynamics which may be affected in the ΔparB strain. In summary, our data demonstrate that ParA is associated with the poles, while the presence of diffuse fluorescence between the poles may indicate a dynamic pattern of localization.
ParA affects the number and positioning of ParB complexes
In most studied bacteria, ParA has been shown to affect the assembly and/or positions of ParB complexes. Our observation that the parA segregation phenotype was suppressed by parB deletion suggested that formation of ParB complexes in the absence of ParA severely disturbs chromosome segregation in Mycobacterium. To investigate this further, we compared the localizations of ParB in the presence and absence of ParA. For this purpose, we constructed strain KG29 (expressing egfp–parA and parB–mcherry) and two strains expressing parB–mcherry from the original locus in the wild-type (KG16) or parA deletion (KG40) backgrounds (Table 1) and verified them as described above. Fluorescent fusions of ParA and ParB were functional, as judged by the low fraction of anucleate or elongated cells for the modified strains (data not shown). Microscopic analysis of the egfp–parA parB–mcherry expressing strain (KG29) revealed that in the majority of cells (80% of 542 cells analysed, Fig. 5A and B), bipolar ParA fluorescence was accompanied by two ParB complexes, which were positioned at 20–25% and 75–80% of the cell length (Fig. 5C). Only a minor fraction (9% of cells) contained only one ParB complex close to or slightly shifted from the cell centre or more than two ParB complexes (3% of cells) (Fig. 5B). In 9% of cells, three ParA complexes were accompanied by more than two ParB foci. The third ParA complex was positioned in the middle of these cells, which were elongated (longer 6 μm), suggesting that they were undergoing division. The increased number of ParB foci and their positions in these cells indicated that a new round of chromosome replication followed by segregation had begun, and ParB complexes in one or both of the daughter cells had already doubled. The pattern of ParB localization in cells of the parB–mcherry strain (KG16) was almost identical to that in cells of the egfp–parA parB–mcherry strain (KG29), indicating that the positioning of ParB complexes was not affected by the expression of EGFP–ParA.
To study the influence of ParA on the localization of ParB, KG40 (ΔparA parB–mcherry; Table 1) cells were studied microscopically. ParB complexes were less intense and more varied in number in cells of the ΔparA parB–mcherry strain (KG40) compared to the parB–mcherry control strain (KG16) (Fig. 5A). Statistical analysis showed that 51.7% of ΔparA parB–mcherry cells (of 204 analysed) contained two ParB complexes, and only 21.2% contained one ParB complex, whereas 27.1% of cells contained three or more ParB complexes (Fig. 5B). Whereas wild-type cells showed a maximum of five ParB complexes, parA deletion cells contained up to seven ParB foci. This difference was not due to a change in the level of ParB–mCherry, since the levels of ParB–mCherry were similar in the ΔparA parB–mcherry (KG40), egfp–parA parB–mcherry (KG29) and parB–mcherry (KG16) strains (Fig. S1). Additionally, the distribution of ParB complexes in the parA deletion strain differed from that observed in KG16 (parB–mcherry) cells. The ParB foci were not confined to positions at 20% and 80% of the cell length (as in the wild type); instead, they were rather randomly distributed along the long axis of the cell (Fig. 5A and C). This pattern was not observed in the control strains, where the ParB foci were always close to the cell axis. Thus, the number and positioning of ParB complexes was found to be highly disturbed in cells lacking ParA.
Mycobacterium smegmatisParA colocalizes with the polar growth determinant Wag31 (DivIVA)
The polar localization of ParA resembled that described previously for the growth determinant Wag31 (a DivIVA homologue, called further DivIVA) in M. smegmatis (Nguyen et al., 2007; Kang et al., 2008), suggesting that the proteins may colocalize. To address this question, we constructed strain KG37, in which vector pMV306pnatdivIVA–mcherry (expressing the divIVA–mcherry gene under the control of the divIVA promoter) was integrated into the chromosome in the KG02 (egfp–parA) background (Table 1). Microscopic analysis of the divIVA–mcherry egfp–parA strain (KG37) revealed that ParA complexes usually overlapped with the DivIVA foci. Interestingly, a large fraction of KG37 cells showed bulging poles and branching. This could be due to the overexpression of divIVA via the additional copy of the divIVA gene (Nguyen et al., 2007). In about half of the cells (48%, 514 cells counted) the fluorescence signals of DivIVA–mCherry and EGFP–ParA were equally distributed between the poles. At the bulged poles of asymmetric cells (52% of cells), however, DivIVA complexes were enlarged and usually accompanied by more intensive ParA fluorescence (Fig. 6). These data suggest that ParA not only colocalizes with DivIVA, but that its localization may be influenced by DivIVA.
ParA interacts with DivIVA
Our microscopic studies demonstrated that ParA and DivIVA colocalize in M. smegmatis cells, leading us to speculate that they could interact. To verify this notion, a bacterial two-hybrid (BTH) system was applied (Karimova et al., 2000). Both N-terminal and C-terminal fusions of M. smegmatis ParA and DivIVA with the T18 and T25 domains of adenylate cyclase (CyaA) were studied. A positive signal was observed only when ParA was N-terminally fused. To pinpoint the interaction interface in DivIVA, an extended BTH analysis was performed using pUT18C and pKT25 constructs containing fragments of the divIVA gene (Fig. 7A and B). Positive signals were obtained for constructs expressing C-terminally truncated (lacking the C-terminal coiled-coil domain IV) and N-terminally truncated (lacking the N-terminal coiled coil) DivIVA, as well as for a version of the protein lacking the variable region between the coiled-coil domains (Fig. 7A). From these results, we conclude that fragments of both coiled-coil regions of DivIVA are involved in the interaction with ParA.
In order to confirm the ParA–DivIVA interaction identified in the BTH studies, pull-down assays were applied. GST–DivIVA was purified from lysates of Escherichia coli BL21 (DE3) cells containing pGEX6P-2divIVA, and then mixed with purified ParA protein (10 μg of protein per extract from 100 ml of culture). To exclude contamination of ParA with GST–ParA and unspecific binding of ParA to the chromatography resin, a control experiment was performed in which ParA was added to lysates from E. coli BL21 (DE3) cells containing the empty pGEX6P-2 vector expressing the GST protein. In an additional control experiment, no ParA protein was added to E. coli BL21 (DE3) pGEX6P-2divIVA cell extracts. Western blotting performed using anti-ParA antibodies detected the protein solely in fractions containing resin-bound GST–DivIVA (Fig. 7C).
In summary, the results of the pull-down assay confirmed the ability of M. smegmatis ParA to interact with DivIVA.
In all bacteria studied to date, the ParA proteins primarily function in active chromosome segregation. However, these proteins are also engaged in other cellular processes, including the regulation of chromosome replication (Murray and Errington, 2008; Scholefield et al., 2012) and, indirectly, cell division (Easter and Gober, 2002). It seems likely, therefore, that these proteins may co-ordinate chromosome segregation with specific cell cycle processes. Here, we investigated this possibility by studying the functions of ParA in Mycobacterium.
Sassetti et al. (2003) suggested that the parA gene may be essential in M. tuberculosis, while Nisa et al. showed that lowering the level of ParA inhibited the growth of M. smegmatis (Nisa et al., 2010). Our present results show that the M. smegmatis parA deletion strain is viable but shows an altered morphology and slower growth, which confirms the importance of ParA for proper cell cycle progression.
In M. smegmatis, we found that elimination of ParA leads to chromosome segregation defects; this is consistent with findings in the other bacteria studied to date, but the fraction of anucleate cells is remarkably high in M. smegmatis cells lacking ParA (30%). Our observation that ParB complexes are mislocalized in the parA deletion strain suggests that the segregation defect is a direct consequence of disturbances in the numbers and positions of ParB complexes. In wild-type cells, ParB forms one or two distinct complexes positioned at 20% and 80% of the cell length. Random positioning of ParB foci within parA-deleted cells suggests that ParA ensures the attachment and/or movement of these foci from the centre to the poles. This function has previously been attributed to ParAs in other bacteria, such as V. cholerae (Fogel and Waldor, 2006) and C. crescentus (Ptacin et al., 2010). Our observation that the elimination of ParA leads to more pronounced segregation defects than those seen with deletion of parB may indicate that either chromosome segregation is more severely perturbed by mislocalized (rather than absent) ParB complexes, or other function(s) of ParA contribute to the observed disorders. Similarly, differences between the segregation phenotypes of the parA and parB deletion strains have been observed in S. coelicolor (26% of ΔparA spores were anucleate versus 17% of ΔparB spores), C. glutamicum (16% of ΔparA cells were anucleate versus 11% of ΔparB cells) (Jakimowicz et al., 2007b; Donovan et al., 2010) and Pseudomonas (Lewis et al., 2002).
Overproduction of ParA leads to the formation of about 10% of anucleate cells, which is similar to the phenotype seen for parB deletion strains. Interestingly, the segregation phenotype is suppressed in parAB deletion mutant, supporting the notion that the presence of mislocalized segrosomes is particularly deleterious. The lack of a segregation phenotype in the parAB deletion strain could also be the result of an elevated copy number of the chromosome, which might ensure random segregation sufficient to provide necessary genetic material to daughter cells. The increased number of chromosomes was detected using the FROS method, in both the parA and the parB deletion strains (I. Santi, pers. comm.). This and the increased number of ParB complexes in the M. smegmatis parA deletion strain may be the result of missegregation. It may also suggest that ParA may be involved in regulating DNA replication, as has been shown for ParA proteins in B. subtilis and V. cholerae (Murray and Errington, 2008; Kadoya et al., 2011). The mechanism of such a regulation could be analogous to that observed in B. subtilis, in which a direct interaction between ParA and the replication initiator protein, DnaA, has been reported (Murray and Errington, 2008; Scholefield et al., 2012). Although no interaction between DnaA and ParA has yet been detected in mycobacteria, the involvement of the ParA protein in the regulation of replication cannot be excluded. In addition to observing an increased number of ParB foci in the parA deletion strain, we also found that their fluorescence intensities were lower. An earlier study (Jakimowicz et al., 2007a) showed that ParA facilitates formation of ParB complexes in vitro. Thus, it is likely that ParB complexes are not properly assembled in the parA deletion strain. Another possible explanation for the weak fluorescence intensity could be a lowered availability of ParB molecules for complex formation.
The other phenotypic defect observed for the parA deletion strain was aberrant septum placement. An influence of ParA on septum positioning has previously been observed in other bacteria, such as S. coelicolor (Jakimowicz et al., 2007b). Aberrant septation may be a direct consequence of chromosome mislocalization and/or missegregation, and the disturbance of ParB complex positioning in the parA deletion background is likely to be connected with altered chromosome organization and/or replication. The presence of a mechanism for co-ordinating septum placement with chromosome positioning could explain this aspect of parA deletion phenotype. In addition, the observed parA overexpression phenotype is consistent with the notion of a direct link between septum placement and segregation, since both septum mispositioning and segregation defects were moderate in parA-overexpressing cells. Until now, however, there has not been any evidence of proteins playing roles equivalent to the Min system or nucleoid occlusion in Mycobacterium (Hett and Rubin, 2008). Since mycobacteria contain other ParA-like proteins, it is also possible that a ParA paralogue which interacts with ParB to regulate cell division (as is the case in C. glutamicum and C. crescentus) becomes misplaced if chromosome segregation fails to proceed properly in the absence of ParA. This would be reminiscent of the function of C. crescentus MipZ which is also a MinD homologue and a ParB interacting protein. MipZ is postulated to be a checkpoint that co-ordinates assembly of the divisome by control of FtsZ polymerization with the onset of DNA replication (Kiekebusch et al., 2012). Another possible, although less likely, explanation for irregular septation is that ParA may exert direct control on septum positioning. The ParA-mediated direct control of septum positioning could lead to the formation of both short (often anucleate) and elongated cells, contributing to the severe segregation phenotype observed for the parA deletion strain. The alteration of cell length observed in parA mutant strains (both deletion and overexpression) may also result from a disturbance in the co-ordination between elongation and the timing of cell division. An influence of ParA on the cell length has also been observed in other bacteria, including C. glutamicum (Donovan et al., 2010) and P. putida (Godfrin-Estevenon et al., 2002), and deletion of parA in S. coelicolor yielded spores of uneven length (Jakimowicz et al., 2007b).
In most M. smegmatis cells, we observed that EGFP–ParA localizes at the poles with the visible patch of fluorescence in between, while longer cells (probably dividing cells) often have an additional complex in the middle of the cell. The localization of ParA suggests that it is associated with the septum during division. A similar septal localization of ParA was previously observed in B. subtilis (Murray and Errington, 2008). The patch of diffuse fluorescence observed between the EGFP–ParA complexes may indicate that ParA has dynamic properties in M. smegmatis cells. Many other ParA homologues, such as those in V. cholerae, C. glutamicum and C. crescentus, exhibit dynamic localization patterns, forming filamentous structures that extend from the pole in a manner dependent on their ATPase activity and interaction with ParB (Fogel and Waldor, 2006; Donovan et al., 2010; Ptacin and Shapiro, 2010). The dynamic properties of M. smegmatis ParA are subject of further analyses.
The polar localization of EGFP–ParA prompted us to speculate that it may interact with a homologue of the pole determinant DivIVA (called Wag31) in M. smegmatis, and our studies confirmed that these proteins colocalize and interact. A previous study showed that DivIVA homologues in actinomycetes are required for polar peptidoglycan sythesis and cell elongation, and overproduction of DivIVA results in overgrowth of one of the cell poles, while its depletion leads to the formation of rounded M. smegmatis cells (Scherr and Nguyen, 2009). An increased amount of ParA was associated with the enlarged DivIVA complexes (possibly due to some disfunction of the DivIVA–mCherry fusion protein), suggesting that localization of ParA depends on DivIVA.
Further studies of DivIVA–ParA interaction suggested that the C-terminal domain of ParA may be responsible for binding to DivIVA (the C-terminal fusions of ParA did not interact in BTH experiments), while the interaction interface in DivIVA is probably formed by the coiled-coil regions flanking the variable third domain, which itself is not required for interaction. Interestingly, the coiled-coil domain II is longer in Mycobacterium DivIVA compared to the other DivIVA homologues, and is followed by a long third domain that contains a phosphorylation site in M. tuberculosis (Nguyen et al., 2007). Phosphorylation of M. tuberculosis DivIVA by PknA/B serine–threonine kinase(s) is growth phase-dependent and may affect the function of DivIVA, as mutations of the phosphorylation site trigger alterations in cell shape (Nguyen et al., 2007). This observation opens the interesting possibility that phosphorylation may affect the affinity of M. smegmatis DivIVA towards ParA by changing the conformation of the interaction interface. The interaction of the segregation machinery with pole determinants has been recently suggested to exist in other actinomycetes. In C. glutamicum, for example, ParB is recruited to the cell poles via an interaction with tip-anchored DivIVA (Donovan et al., 2012), while in S. coelicolor, ParA interacts with another tip-associated protein, Scy (B. Ditkowski, submitted). The tip anchorage of ParA is not specific to actinomycetes; it has also been identified in C. crescentus, mediated through association with the polar protein TipN (Ptacin et al., 2010). Interestingly, the pole-associated DivIVA of B. subtilis interacts with MinD, which belongs to the same family as ParA (Marston et al., 1998). The identified interaction between ParA and DivIVA in M. smegmatis may therefore contribute to the complex phenotype of parA deletion strain, especially the observed alterations in cell length. This notion critically contributes to the growing body of evidence that bacterial ParA homologues are associated with cell division and/or cell elongation.
In summary, deletion of parA in M. smegmatis yields a complex phenotype that may be associated with the multifunctional segregation machinery and is accompanied by disturbed cell division/elongation control. A presumed similar function of ParA in co-ordinating the cell cycle in M. tuberculosis and the fact that it is postulated to be essential together suggest that ParA and (more precisely) the ParA–DivIVA interaction may be potential drug targets for the treatment of tuberculosis.
DNA manipulation, bacterial strains and growth conditions
DNA manipulations were carried out using standard protocols (Sambrook et al., 1989). Enzymes were supplied by Fermentas, and oligonucleotides were obtained from Genomed. All PCR-derived clones were confirmed by DNA sequencing. E. coli BL21 (DE3) was used as the host for overproduction of fusion proteins. E. coli strains were grown in Luria–Bertani (LB) medium at 37°C. M. smegmatis mc2 155 and its derivatives were grown in Middlebrook 7H9 medium and on 7H10 agar plates supplemented with OADC (oleic acid, albumin, glucose, sodium chloride; Difco), in NB broth [8.0 g l−1 nutrient broth (Difco), 10.0 g l−1 glucose and 0.2% Tween 80, pH 6.0–6.2] or on NB agar plates. For selection, kanamycin (25 mg ml−1) was used as needed. Acetamide (final concentration 0.1%) was used to induce ParA overproduction in KG11 cells (M. smegmatis mc2 pMV306pamiparA).
Construction of mutant strains in the M. smegmatis mc2 155 background
To perform unmarked deletions and insertions in the M. smegmatis mc2 chromosome, targeted gene replacement was performed according to the protocol of Parish and Stoker (2000). The constructs, which were based on vectors p2Nil (KanR) and pGOAL17 (sacB, lacZ) (Parish and Stoker, 2000), were prepared according to the strategy described in Supporting information and integrated into the M. smegmatis chromosome by homologous recombination. For transformation of electrocompetent M. smegmatis cells, the plasmid DNA was treated with NaOH/EDTA (0.2 mM/0.2 mM). Transformants were plated on NB plates and selected for kanamycin resistance. The KanR SCO (single-crossover recombinant) colonies were blue and sensitive to sucrose (2%). The SCO colonies were further plated on NB without selection. Cells were then resuspended in liquid medium and serial dilutions were plated onto NB plates supplemented with sucrose and X-Gal. The selected double-crossover (DCO) mutants were white, KanS and resistant to sucrose.
PCR and Southern hybridization were used to distinguish between wild-type and DCO mutant cells. The desired deletions and/or fusion proteins were confirmed by Western blotting analyses.
For construction of M. smegmatis complementation or meroploid strains, derivatives of the pJAM2 shuttle vector (Triccas et al., 1998) or the mycobacteriophage L5-based integration-proficient vector pMV306 (Murry et al., 2005) were used (see Supporting information for details on their construction). Vectors were introduced into M. smegmatis-competent cells by electroporation, and transformants were selected using kanamycin.
In all strains, the presence of plasmid DNA was confirmed by PCR analysis of DNA isolated from the mutant strains, as described previously (Dziadek et al., 2003). The protein levels in cell extracts were analysed by SDS-PAGE (100 μg of cell extracts were loaded on the gel) followed by Western blotting performed using polyclonal anti-ParA or anti-ParB according to standard procedures (Towbin et al., 1979). The presence of fluorescent fusion proteins was verified by scanning the fluorescence in the SDS-PAGE gel as described before (Jakimowicz et al., 2005). The level of parA overexpression in KG11 was assessed by immunoblotting different amounts of KG11 and control cells, and quantifying the results (ImageQuant software, Molecular Dynamics).
Mycobacterium smegmatis wild-type, KG22, KG23, KG11, KG03 and ΔparB (Jakimowicz et al., 2007a) strains were cultured in rich (7H9 Middlebrook supplemented with OADC) medium. Starter cultures were grown overnight in rich medium, and growth assays were conducted with an initial OD600 of 0.05. For overproduction of ParA, 0.1% acetamide was added; the M. smegmatis mc2 pMV306 strain served as a wild-type control for acetamide induction. Cells were incubated with shaking for 24 h, and samples were collected every 3 h for optical density analysis and CFU determination. Assays were performed at least in triplicate.
Mycobacterium smegmatis cells (log-phase cultures grown in 7H9 Middlebrook supplemented with OADC) were washed and resuspended in phosphate-buffered saline (PBS; 10 mM sodium phosphate, pH 7.4, 150 mM NaCl, 15 mM KCl). Nascent peptidoglycan synthesis labelling was performed as previously described (Daniel and Errington, 2003). Briefly, BODIPY FL-conjugated vancomycin (Invitrogen; final concentration 2 μg ml−1) mixed with an equal amount of unlabeled vancomycin was added to growing cultures, which were then incubated with shaking for 3 h to allow binding of the labelled antibiotic within the cell wall. Cells were viewed directly by microscopy. For analysis of chromosome segregation and elongation defects, the cells were additionally permeabilized by exposure to toluene (2%) for 2–5 min, washed, resuspended in PBS and stained with DAPI (2 μg ml−1) for 30 min at room temperature. The cells were then incubated for 5 min in 1.5 μg ml−1 FM4-64 for visualization of membranes, and examined under a Zeiss Axio Imager Z1 equipped with an X100 objective.
Purification of mycobacterial ParA and DivIVA proteins
To overproduce recombinant GST–ParA and GST–DivIVA fusion proteins in M. smegmatis, the pGEX-6P-2 vector was used. The parA and divIVA genes were PCR amplified using appropriate primers (pJAMAMsF and pJAMAMsR for parA; DivIVMsF and DivIVMsR for divIVA) and chromosomal DNA as the template. The resulting PCR products were cloned into the pGEM-T Easy vector (Promega) and recloned into the BamHI/NotI (parA gene) or BamHI/EcoRI (divIVA gene) sites of pGEX-6P-2 (GE Healthcare). The glutathione S-transferase fusion proteins were purified on GSH-Sepharose 4B columns (GE Healthcare) and treated with the PreScission protease (GE Healthcare), as previously described (Jakimowicz et al., 2007a). The purified proteins were more than 95% homogeneous, as assessed by SDS-PAGE analysis.
A modified affinity chromatography procedure was used for the pull-down assays. Briefly, 10 μg of ParA and GSH-Sepharose resin was added to 100 ml of E. coli BL21 lysate (suspended in 50 mM Tris, pH 8.0, 150 mM NaCl and 1 mM ATP) containing overproduced GST–DivIVA. After incubation for 2 h at 4°C, the GSH-Sepharose beads were extensively washed with buffer containing 50 mM Tris, pH 8.0, and 150 mM NaCl. Bound protein complexes were eluted with glutathione solution (20 mM glutathione, 100 mM Tris, pH 8.0, and 100 mM NaCl) and analysed by SDS-PAGE and Western blotting with anti-MsParA.
Bacterial two-hybrid (BTH) assays
The recombinant plasmids used in the BTH system (Karimova et al., 2000) were constructed by PCR amplification of the parA and divIVA genes using appropriate primers and chromosomal DNA templates (Table S1). The amplified genes were cloned into the appropriate sites of the pUT18C and pKT25, or pUT18 and pKNT25 vectors. The resulting recombinant plasmids expressed the proteins of interest fused to the C- or N-terminus (respectively) of the T18 and T25 fragments of adenylate cyclase. For BTH assays, recombinant or empty pKT25 and pUT18C plasmids were used in various combinations to co-transform BTH101 cells. The transformants were plated onto LB/X-Gal medium and incubated at 30°C for 48 h.
This work was supported by the Wroclaw Research Center EIT+ under the project Biotechnologies and advanced medical technologies – BioMed (POIG 1.1. Project 3.1.) financed from the European Regional Development Fund (Operational Program Innovative Economy, 1.1.2).
We are very grateful to John McKinney and Isabella Santi for critical comments on the manuscript.