|Home | About | Journals | Submit | Contact Us | Français|
FtsZ, a bacterial homolog of tubulin, forms a structural element called the FtsZ ring (Z ring) at the predivisional midcell site and sets up a scaffold for the assembly of other cell division proteins. The genetic aspects of FtsZ-catalyzed cell division and its assembly dynamics in Mycobacterium tuberculosis are unknown. Here, with an M. tuberculosis strain containing FtsZTB tagged with green fluorescent protein as the sole source of FtsZ, we examined FtsZ structures under various growth conditions. We found that midcell Z rings are present in approximately 11% of actively growing cells, suggesting that the low frequency of Z rings is reflective of their slow growth rate. Next, we showed that SRI-3072, a reported FtsZTB inhibitor, disrupted Z-ring assembly and inhibited cell division and growth of M. tuberculosis. We also showed that M. tuberculosis cells grown in macrophages are filamentous and that only a small fraction had midcell Z rings. The majority of filamentous cells contained nonring, spiral-like FtsZ structures along their entire length. The levels of FtsZ in bacteria grown in macrophages or in broth were comparable, suggesting that Z-ring formation at midcell sites was compromised during intracellular growth. Our results suggest that the intraphagosomal milieu alters the expression of M. tuberculosis genes affecting Z-ring formation and thereby cell division.
Mycobacterium tuberculosis, the causative agent of tuberculosis, is an important infectious agent that globally causes more than three million new infections each year (8). Recent years have seen an increase in the number of M. tuberculosis strains that are resistant to one or more antituberculosis drugs, and this has highlighted the need for the development of a new generation of antimicrobial agents. One hallmark of the M. tuberculosis life cycle is that it exists in two metabolically distinct growth states: an active replicative state and a nonproliferative persistent state where the bacterium survives without any increase in the bacterial burden on the host. Physiological studies carried out by Wayne and colleagues indicate that M. tuberculosis cells in the hypoxia-induced nonreplicative persistent state are blocked at the cell division stage after completing DNA replication and undergo a round of cell division prior to initiation of a new round of DNA replication (40, 41). This latter process is also referred to as reactivation. Development of antimycobacterial agents targeting the cell division process could potentially prevent the multiplication and subsequent proliferation of the pathogen in active, as well as reactivation, growth states.
FtsZ, a bacterial homolog of tubulin, is a key player in cell division and is essential for initiation of this process (22, 32). FtsZ protein catalyzes the formation of distinct structures, referred to as FtsZ rings (Z rings), at the midcell site and sets up a scaffold for ordered assembly of other cell division proteins. The combined action of multiple cell division proteins results in septation (22, 32). FtsZ protein-catalyzed Z-ring assembly represents the earliest known step in the septation process. FtsZ protein polymerizes in vitro into protofilaments in a GTP-dependent manner, and its assembly dynamics are regulated by GTP hydrolysis (25). FtsZ is a well-conserved protein that is present in nearly all prokaryotes (22). Due to its central and essential role in bacterial cytokinesis, and its absence in higher eukaryotes, the FtsZ protein is considered an attractive antimicrobial drug target (3, 19, 21, 22, 25, 44).
Earlier studies on ftsZ and the cell division process in mycobacteria focused on Mycobacterium smegmatis, a rapid grower with an average doubling time of 3 h. These studies indicated that ftsZ is an essential cell division gene (10) and that M. tuberculosis is exquisitely sensitive to the intracellular levels of FtsZ (FtsZTB), as constructs expressing ftsZTB from native or heterologous promoters are not stably maintained (9). Because of the toxicity associated with elevated expression levels of ftsZ in M. tuberculosis, attempts to visualize FtsZ structures in M. tuberculosis have not been successful. At the biochemical level, FtsZTB has been purified, characterized, and found to exhibit slow polymerization and weak GTPase activities in vitro (30, 43).
We have been unable to localize FtsZ structures in mycobacteria by immunohistochemistry due, perhaps, to their thick and unyielding cell walls (9). This feature, combined with the toxicity associated with the elevated levels of ftsZTB expression in M. tuberculosis, led us to develop an ftsZTB reporter strain where FtsZ-green fluorescent protein (GFP) fusion protein can function as the sole source of FtsZ (10). With this strain, we visualized FtsZTB structures in M. tuberculosis grown under different conditions. We describe here the FtsZ localization in cells growing in culture and in macrophages.
Escherichia coli Top10, used for cloning, was propagated in Luria-Bertani broth, and transformants were selected on Luria-Bertani agar containing either kanamycin (Km, 50 μg/ml) or hygromycin (Hyg, 50 μg/ml). M. smegmatis mc2-155 and M. tuberculosis H37Ra (and H37Rv) were grown in Middlebrook 7H9 broth supplemented with oleic acid, albumin, dextrose, and sodium chloride. Transformants were selected in the same medium supplemented with agar containing Km (10 μg/ml), Hyg (50 μg/ml), or both (10). As needed, acetamide was supplied in growth medium at a final concentration of 0.2%. In some experiments, actively growing cultures of M. tuberculosis were exposed to the small-molecule inhibitor SRI-3072 at 0.5 μM (about equal to the MIC). Growth was followed for several days after exposure by monitoring absorbance at 600 nm, and viability was measured by determining CFU.
Plasmids pACR1 and pJFR66 were created by cloning the PCR-amplified fragments encompassing the ftsZsmeg and ftsZTB coding regions and their respective 1-kb 5′ flanking regions (Table (Table1)1) in integrating plasmids. Plasmid pJFR41 was created by cloning the ftsZTB-gfp fusion (9) downstream of the amidase promoter in pJFR19 (Table (Table1).1). The gfp gene in pJFR41 was derived from the fluorescence-activated cell sorter-optimized mut3 variant amplified from pFV25 (5, 9). All PCR products were confirmed by sequencing.
A suicide recombination plasmid, pJFR52, containing the 3.6-kb ftsZTB gene region with an 840-bp internal deletion in the ftsZTB gene was constructed in two steps. First, a 2.1-kb DNA fragment bearing the 5′ end of ftsZTB and its upstream flanking region and a 1.6-kb fragment bearing the 3′ end of ftsZTB and its downstream flanking region, were amplified by using oligonucleotide primer pair MVM276 and MVM238 and pair MVM280 and MVM281, respectively (Table (Table2).2). The resultant PCR products were cloned adjacent to each other in vector p2NIL to create pJFR51 (27). Next, a 6.1-kb PacI fragment carrying the lacZ, aph, and sacB genes was isolated from pGOAL17 and inserted into pJFR51 to create suicide recombination plasmid pJFR52 (27).
The pJFR52 plasmid was electroporated into M. tuberculosis H37Ra, and single crossovers were selected on agar plates containing Km and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside. A blue, Km-resistant single-crossover strain, M. tuberculosis 52, was confirmed by PCR. To inactivate ftsZ at its native location, a plasmid construct (pJFR66 in Table Table1;1; Fig. Fig.1)1) expressing ftsZ from its native promoter was integrated at the bacteriophage attachment site of M. tuberculosis 52, and the resultant merodiploid strain was screened for double crossovers (DCOs) as previously described (10). White, Km-sensitive, Hygr, and sucrose-resistant DCO colonies were analyzed by PCR and Southern hybridization with ftsZTB-specific, 32P-labeled probes. One strain, designated M. tuberculosis 66, was confirmed to be chromosomally null for ftsZ and contained an integrated copy of PftsZ-ftsZTB. Integration at the attB site in mycobacteria can be efficiently excised by phage excisionase and replaced simultaneously with the incoming plasmid carrying alternate antibiotic markers (28). With this strategy, we switched the integrated plasmid expressing ftsZ from its native promoter (pJFR66, Hygr) by transforming the M. tuberculosis 66 strain with a KM resistance-encoding incoming plasmid expressing PftsZ-ftsZsmeg, pACR1 (Table (Table1),1), to generate M. tuberculosis ACR1. Next, we swapped the resident plasmid in M. tuberculosis ACR1 with pJFR41 (Pami-ftsZTB-gfp, Hygr) to create M. tuberculosis 41. Inclusion of the pACR1 swapping step was necessary to generate M. tuberculosis 41, as both pJFR66 and pJFR41 carried a gene for Hygr, which would have made the screening process cumbersome. Since pACR1 expressed M. smegmatis ftsZ, these results also indicated that the M. smegmatis counterpart could substitute ftsZTB function. Transformants with pJFR41 were plated on agar containing 0.2% acetamide. The DCO strains were confirmed by PCR amplification and sequencing of the integrated copy of the gene and by Southern hybridization with ftsZTB and gfp gene-specific probes.
M. tuberculosis genomic DNA was isolated from various strains, digested with XhoI, and processed for Southern hybridization as previously described (33). Nitrocellulose blots were hybridized with PCR-generated, 32P-labeled ftsZTB (Table (Table2)2) and gfp (Table (Table2)2) probes (10).
Immunoblotting was carried out to detect FtsZTB and FtsZTB-GFP in cellular lysates of broth- and in vivo-grown wild-type M. tuberculosis and M. tuberculosis 41 as previously described (10). We used M. tuberculosis SigA protein to normalize for protein amounts loaded per lane when comparing the FtsZ levels in broth- and macrophage-grown M. tuberculosis. SigA levels are not known to change during intracellular growth of M. tuberculosis (45). Blots were probed simultaneously with anti-FtsZTB antibodies and monoclonal antibodies to the sigma 70 subunit of E. coli RNA polymerase. The latter have been shown to bind mycobacterial SigA protein (29, 45). Anti-sigma 70 antibodies were obtained from Neoclone Biotechnology (Madison, WI) and used as recommended. Immunoblots were processed with the ECF Western blotting kit from Amersham (Piscataway, NJ) and scanned on a Bio-Rad Molecular Imager (FX), and FtsZ levels were determined with the volume analysis function of the QuantityOne software.
Wild-type M. tuberculosis and M. tuberculosis 41 were grown for various periods of time with shaking, harvested by centrifugation, washed in phosphate-buffered saline, fixed in 1% paraformaldehyde, and stored at 4°C until further use. Bacteria were examined by bright-field and fluorescence microscopy with a Nikon Eclipse 600 microscope equipped with a 100× Nikon Plan Fluor oil immersion objective with a numerical aperture of 1.4 and a standard fluorescein isothiocyanate filter set (Chroma). Images were acquired with a Photometrics Coolsnap ES camera and Metapmorph 6.2 imaging software (Universal Imaging Corporation). Images were optimized with Adobe Photoshop 7.0. Some images were processed with the homomorphic fast Fourier transform (FFT) filtering function of the Metamorph 6.2 software. When applied to an image, this function performs simultaneous contrast enhancement and compression of the brightness dynamic range.
Monocyte-derived human macrophage cell line THP-1 was infected with either M. tuberculosis or M. tuberculosis 41. Uninfected THP-1 cells were maintained in RPMI medium with 10% fetal bovine serum. Prior to infection, THP-1 cells were exposed to 50 nM phorbol-12-myristate-13-acetate for 24 h and allowed to differentiate into macrophages. Approximately 5 × 105 cells/ml were infected with M. tuberculosis or M. tuberculosis 41 at a multiplicity of infection of 1:10 (macrophage/bacterium ratio). After 3 h of phagocytosis, macrophages were washed to remove nonphagocytosed bacteria and further incubated. At the indicated time points, either the macrophages were lysed with 0.09% sodium dodecyl sulfate (SDS) and bacteria recovered following centrifugation at 14,000 rpm for 5 min or the infected macrophage monolayers were washed three times with phosphate-buffered saline, scraped, and resuspended in Tris-EDTA buffer. The recovered bacteria or macrophages containing bacteria were lysed by bead beating for 3 min in a mini bead beater. Cleared lysates were obtained by centrifugation, separated on SDS-polyacrylamide (PA), transferred to nitrocellulose, and probed for FtsZ levels as outlined above. For microscopy, the recovered bacteria were fixed in 1% paraformaldehyde and visualized by bright-field or fluorescence microscopy, as needed.
Our approach to study FtsZTB-mediated cell division in M. tuberculosis is to construct an ftsZ-gfp mutant strain and investigate FtsZTB ring assembly under different growth conditions. Our earlier studies revealed that self-replicating plasmid constructs expressing ftsZTB in M. tuberculosis from either native or heterologous promoters are unstable (9). Also, intense fluorescent FtsZTB structures in M. smegmatis merodiploids can be visualized if FtsZTB is produced from the amidase promoter (Pami-ftsZTB-gfp) but not from its native promoter (PftsZ-ftsZTB-gfp) (9, 30). Bearing these two data in mind, we constructed an ftsZTB reporter strain in which FtsZTB tagged with GFP was the sole source of FtsZ (see below). Once constructed, the strain was characterized with respect to FtsZ levels, FtsZ structures, and growth in broth and macrophages.
We used the two-step recombination protocol of Parish and Stoker to disrupt the native ftsZ gene in the presence of an integrated copy of ftsZTB (27). Mapping of ftsZTB transcriptional start points identified four promoters, with the farthest one at 787 nucleotides upstream of the ftsZTB start codon (data not shown). Accordingly, a DNA fragment bearing the ftsZTB coding region and its 1-kb upstream flanking region was amplified, cloned in integrating vector pMV306H (pJFR66 in Table Table1),1), and used during the selection of DCOs as described in Materials and Methods. One mutant DCO, designated M. tuberculosis 66 and carrying a functional copy of ftsZTB at the attB site, was selected and used as the base strain to generate the ftsZTB-gfp reporter strain (M. tuberculosis 41) by a plasmid-swapping protocol (Fig. (Fig.11 and Materials and Methods) (28, 30). Southern hybridization of M. tuberculosis 41 genomic DNA with the ftsZ gene probe identified two bands: one corresponding to the integrated ftsZ copy and the other to the mutant copy carrying an 840-bp internal deletion in the ftsZTB gene (Fig. (Fig.2A).2A). A parallel blot hybridized with the gfp gene probe identified only one band corresponding to the integrated copy (Fig. (Fig.2B).2B). These results confirmed that the transformation-based plasmid-switching protocols successfully replaced the resident plasmid carrying PftsZ-ftsZTB with an incoming plasmid containing Pami-ftsZTB-gfp.
To further validate the Southern data, M. tuberculosis 41 grown in the presence of 0.2% acetamide was examined for FtsZTB-GFP production by immunoblotting with anti-FtsZTB and anti-GFP antibodies. When probed with anti-FtsZTB antibodies, a distinct band corresponding to FtsZ-GFP fusion protein (65 kDa) and none corresponding to FtsZTB (39 kDa) was detected in M. tuberculosis 41 lysates (Fig. (Fig.2C),2C), whereas a 39-kDa band corresponding to FtsZTB was detected in the lysates of wild-type M. tuberculosis (Fig. (Fig.2C).2C). A parallel blot probed with anti-GFP antibodies revealed only one band corresponding to FtsZTB-GFP in M. tuberculosis 41 lysates and none in the parent strain M. tuberculosis (Fig. (Fig.2D,2D, compare lane 2 with lane 1). Quantification of FtsZ and FtsZ-GFP bands in Fig. Fig.2C2C revealed that the levels of FtsZ-GFP fusion protein in M. tuberculosis 41 were comparable to those of the native protein produced in the parent strain (the ratio of FtsZ-GFP to FtsZ was 0.9:1) (Fig. (Fig.2C).2C). It is interesting that expression of ftsZTB-gfp from the amidase promoter in self-replicating plasmids in M. smegmatis resulted in the accumulation of excess fusion protein (12, 15). Thus, the nearly normal levels of FtsZTB-GFP in M. tuberculosis 41 cells suggest that FtsZTB levels in M. tuberculosis are more tightly regulated than in M. smegmatis.
The viability of M. tuberculosis 41 decreased by nearly 5 log units when actively growing cultures were plated on medium lacking acetamide (Fig. (Fig.3A).3A). The growth rate of M. tuberculosis 41, slightly slower than that of wild-type M. tuberculosis, slowed down further in the absence of acetamide (Fig. (Fig.3B).3B). Although immunoblotting did not reveal any significant differences in FtsZ-GFP levels when the strain was grown with and without acetamide for four doublings (data not shown), the absence of inducer led to a 20% increase in average cell length (from 2.47 μm [n = 119] to 2.98 μm [n = 105]). Thus, growth in the absence of acetamide inhibited cell division and led to a reduction in the viability of M. tuberculosis 41. Therefore, loss of viability perhaps occurs before major changes in the FtsZ level become apparent. Furthermore, immunoblotting may not be sensitive enough to discern the small changes in FtsZ levels that are nevertheless able to affect the cell division of M. tuberculosis 41 grown in the absence of acetamide. Expression from the inducible amidase promoter is known to be leaky in M. tuberculosis (4, 9, 10, 15; our unpublished data). Since M. tuberculosis 41 required acetamide for viability, these data also suggest that the leaky expression is not sufficient to sustain the growth of this strain. Furthermore, growth in the absence of acetamide beyond four doublings may be required to see a reduction in FtsZ levels. Together, the above results confirm that FtsZTB-GFP is the only FtsZ protein produced in M. tuberculosis 41 and suggest that it is functional in M. tuberculosis cell division. It is pertinent to note that although merodiploid strains producing FtsZ-GFP fusion proteins have been reported in other bacteria, efforts to utilize an ftsZ reporter strain where ftsZ-gfp functions as the sole source of ftsZ have met with limited success. For example, in E. coli, where ftsZ dynamics are well characterized at the genetic and biochemical levels, FtsZ-GFP is not fully capable of replacing the function of native FtsZ (20, 37). Similarly, fusion of the only copy of ftsZ to gfp in Bacillus subtilis resulted in a temperature-sensitive phenotype due, perhaps, to the inability of the fusion protein to fold properly at high temperature (17). In Streptomyces coelicolor, ftsZ-gfp is capable of complementing an ftsZ chromosomal null mutation but the resultant strain exhibits a delayed and defective sporulation phenotype (14).
Next, we visualized FtsZTB-GFP structures by fluorescence microscopy in actively growing cells of M. tuberculosis 41. The majority of cells with FtsZ-GFP structures had distinct midcell FtsZ bands (Fig. (Fig.4),4), although some cells had FtsZ localized at poles (arrows in Fig. 4a and c). Sometimes, cells with midcell Z rings showed faint polar fluorescence, and conversely, cells with distinct polar spots showed faint or incomplete Z rings. Approximately 11% of actively growing cells had midcell Z rings (Table (Table3).3). M. tuberculosis is a slow grower with an average doubling time of 24 h. Thus, the observed low frequency of midcell Z rings presumably reflects the actual percentage of cells undergoing cell division. It is pertinent to note that approximately 30% of actively growing M. smegmatis ftsZ merodiploid cells contain midcell Z rings (30, 31). Presumably, the frequency of Z-ring formation is proportional to the growth rate of mycobacteria. Growth rate-dependent changes in the frequency of medial Z rings have been reported for B. subtilis (18, 42). Recently, Erickson and colleagues measured fluorescence recovery after photobleaching of GFP-tagged FtsZ proteins of E. coli and B. subtilis and concluded that the Z ring is highly dynamic and continuously remodels itself with a half-time of 8 s (1, 35). It will be interesting to determine whether the Z-ring assembly dynamics in slow growers such as M. tuberculosis are proportionately slower.
The average length of M. tuberculosis 41 cells with distinct polar structures was ~2.2 μm (N = 38), whereas that of the cells with evident midcell bands was 4.0 μm (N = 59). Since wild-type M. tuberculosis cells grown under similar conditions were ~2.1 μm (N = 100) in length, this approximately twofold increase suggested that the polar structures could be remnants of septa from the previous division event. The above interpretation assumes that the polar localizations of FtsZ were not unique to M. tuberculosis 41 and could be observed in the parent strain. Alternatively, it is possible that interactions of FtsZTB and negative regulators of Z-ring assembly in M. tuberculosis, if any, were perturbed in M. tuberculosis 41, resulting in localization of FtsZ at non-midcell sites. We tend to favor the first interpretation because the average size of cells with polar localizations was similar to the average length of actively growing M. tuberculosis cells. Most cells showed dark coloration at the cell poles. While the exact nature of these dark spots is unclear, they could be due to the external ridges observed at the cell poles of M. tuberculosis by transmission electron microscopy (6).
Recently, a group of structurally diverse small-molecule inhibitors, named zantrins, was shown to perturb the Z-ring assembly in E. coli and inhibit the growth of several bacterial species in broth cultures. These compounds interfered with the GTPase activity of E. coli FtsZ (FtsZEC) and FtsZTB, caused destabilization of FtsZEC protofilaments, increased filament stability, and in some cases interfered with Z-ring assembly (21). The effects of zantrins on M. tuberculosis growth and FtsZTB assembly were not examined in these experiments. We (R.R.) recently showed that SRI-3072, a small-molecule inhibitor belonging to a class of 2-alkoxycarbonylaminopyridines, inhibited the growth of M. tuberculosis with an MIC of 250 ng/ml (0.47 μM) (44). This compound also inhibited the GTPase activity of FtsZTB in vitro, albeit with low affinity (i.e., 20% reduction in activity at 100 μM). Since it was unknown whether SRI-3072 affected FtsZ polymerization and Z-ring assembly in vivo, we addressed this question with M. tuberculosis 41.
Actively growing cultures of M. tuberculosis 41 were exposed to 0.56 μM SRI-3072 for various times, and effects of the inhibitor on growth and FtsZTB structures were examined. As expected, SRI-3072 interfered with the growth of M. tuberculosis 41 (Fig. (Fig.5A).5A). Fluorescence microscopy revealed a gradual disappearance of Z rings (Fig. (Fig.5B,5B, parts a, c, e, and g) with increasing times of exposure. After 24 h of exposure, a reduction in the number of cells containing midcell Z rings was noted, although FtsZ-GFP localization at random spots was evident (data not shown; Fig. Fig.5B,5B, parts e to h). After 48 h of exposure, FtsZ-GFP localization at random spots also became compromised and a small increase in cell length was noted (Fig. 5B and C). Midcell FtsZTB-GFP bands were present in approximately 2.2% of drug-treated cells, whereas they accounted for 11% in untreated controls (Table (Table3).3). By day 5, almost no distinct Z rings were evident; rather, only diffuse and faint fluorescence was seen in most cells (Fig. (Fig.5B,5B, parts g and h). A modest increase in cell length combined with the disappearance of Z rings is consistent with the interpretation that SRI-3072 interfered with FtsZTB ring assembly and cell division. It is pertinent to note that zantrins, which inhibit the growth of a wide range of bacteria, did not cause overt filamentation (21). In comparison to SRI-3072-treated cells, 4% of untreated cells had midcell Z rings after 120 h of growth (not shown). We have shown previously that the FtsZ levels in M. tuberculosis decrease during the stationary phase (9). The reduction in the number of cells with midcell Z rings at 120 h of growth presumably reflects the fact that these cells were in the stationary phase of growth. Treatment of M. tuberculosis 41 with SRI-3072 for 72 h caused an approximately 33% decrease in FtsZ levels, whereas no change in FtsZ levels was noted in untreated controls (data not shown). Interestingly, removal of SRI-3072 after 48 h of exposure did not result in recovery of viability (data not shown). It is possible that the compound SRI-3072 has inhibitory effects on other metabolic processes as well. The development and characterization of new antimycobacterial agents that affect M. tuberculosis proliferation are of great importance. The M. tuberculosis 41 reporter strain can potentially be used for evaluating the effects of new and reported inhibitors of FtsZTB activities in vivo.
M. tuberculosis multiplication inside eukaryotic host cells is critical for virulence. It is unknown whether FtsZ assembly and cell division are affected during growth in macrophages. To begin addressing this issue, human macrophage-derived monocyte cell line THP-1 was infected with M. tuberculosis 41 and cultured for 3 days. We then recovered the intracellular bacteria and visualized them by fluorescence microscopy (Fig. (Fig.6).6). Results were compared with the cultures grown in broth. Two observations were readily apparent. First, macrophage-grown M. tuberculosis 41 cells were filamentous (Fig. (Fig.6C;6C; compare parts i and ii with iii and iv) compared to broth-grown cultures, strongly suggesting a defect in cell division. The wild-type M. tuberculosis strain also showed filamentation upon growth in THP-1 (Fig. 6A and B), indicating that the cell elongation phenotype is a characteristic feature of intracellular M. tuberculosis. Some filamentous cells also contained buds and bulges (Fig. (Fig.6C,6C, arrowheads in parts iii and iv). Such structures were reported for M. smegmatis grown under conditions that increase FtsZ (10) or deplete WhmD (12). Second, fluorescence microscopy revealed that, in contrast to broth-grown cultures, a majority of macrophage-grown M. tuberculosis 41 cells had several nonring structures along the entire length of the cell, and only a small population of filamentous cells (1 to 3%) contained distinct Z rings at midcell sites (Fig. (Fig.6C,6C, part iii). Processing of fluorescent images by the homomorphic FFT filtering function of the Metamorph 6.2 software revealed diffuse, spiral-like structures (Fig. (Fig.6D,6D, parts i to vi). Because of the narrow width of M. tuberculosis cells, we did not succeed in improving the quality of these images to better discern the spiral structures. Since M. tuberculosis cells continue to proliferate in macrophages, we reasoned that these diffuse spiral structures were intermediates in FtsZ assembly (38) and would eventually lead to productive Z rings and subsequent cell division.
Localization of FtsZ in nonring structures has also been observed under FtsZ overproduction conditions in E. coli (20) and during sporulation in B. subtilis (2). Studies based on fluorescence recovery after photobleaching indicated that only 30% of the total FtsZ in E. coli is in the Z ring, with the rest in the cytoplasm (36). Time-lapse analysis of FtsZ structures indicated that FtsZ outside of the Z ring is in highly dynamic, potentially helical cytoskeletal structures (38). It was suggested that the highly mobile structures serve to scan the cell surface for potential division sites more efficiently. It remains to be established whether the diffuse, spiral-like structures observed during the intracellular growth of M. tuberculosis 41 cells are comparable to the well-characterized FtsZ structures of E. coli (38) and B. subtilis (2).
We considered whether the low frequency of Z rings at midcell sites during intracellular growth was due to altered levels of FtsZ. Cellular lysates from macrophage-grown bacteria were prepared and FtsZ levels were determined by immunoblotting with anti-FtsZTB antibodies. Since protein lysates prepared from these bacteria could be contaminated with small amounts of macrophage proteins, FtsZ levels were normalized to those of SigA. The levels of SigA, a housekeeping sigma factor, are known to be stably maintained under various conditions of growth in broth and in vivo (13, 45). The immunoblots were therefore probed simultaneously with anti-FtsZTB and monoclonal anti-σ70 antibodies. Analysis by a fluorescence imager indicated that the ratio of FtsZ to SigA in lysates prepared from bacteria grown in macrophages was comparable to the ratio obtained for broth-grown bacteria (Fig. (Fig.7;7; data not shown).
Together, the above results suggested that the altered activities of FtsZ, and not the altered protein levels, were responsible for the observed nonring structures during intracellular growth. This raises a question as to why such nonring FtsZ structures were abundant during intracellular growth but were not readily detectable during growth in broth. We propose that the FtsZTB assembly at the midcell site is regulated by a hitherto unidentified accessory factor(s) whose activity could be compromised or altered during intracellular growth, thereby resulting in diffuse nonring structures and filamentation. Based on the genome data, M. tuberculosis appears to lack orthologs of known regulators and stabilizers of Z-ring and FtsZ-interacting proteins. To date, the only interactions of FtsZTB reported are those with FtsWTB (7). Although the FtsZ-FtsW interaction is critical for cell division, FtsZ can localize to the midcell site independently of FtsW (31). The intracellular environment that M. tuberculosis faces upon infection is believed to be hostile and rich in reactive nitrogen and oxygen intermediates, cytokines, and antimicrobial peptides. It is also acidic and hypoxic in nature and nutrient limited (34). M. tuberculosis adapts to the stressful intracellular environment by modulating the expression (34) of a wide array of genes, including perhaps those responsible for the observed diffuse FtsZ structures and filamentation. Presumably, the balance of proteins promoting and inhibiting Z-ring assembly is perturbed during growth of M. tuberculosis in macrophages. However, an alternate possibility, that FtsZ-GFP fusion protein is less stable and that filamentation caused by hitherto unknown mechanisms during intramacrophage growth readily disassembles midcell FtsZ-GFP rings, remains open. It should be noted that FtsZ-GFP spiral-like structures were evident during intramacrophage growth, although there was a reduction in the number of midcell FtsZ-GFP rings (Fig. (Fig.6).6). It is likely that the stability of FtsZ-GFP in spiral-like structures is different from that in midcell rings.
Bacterial filamentation is often triggered by a wide variety of factors, including exposure to DNA-damaging agents and to antibacterial agents that interfere with FtsI activity (recently reviewed in reference 24). Filamentation during intracellular growth has also been reported for some gram-negative pathogens. For example, Salmonella enterica serovar Typhimurium growing in murine fibroblast cells (23) and contractile vacuoles of amoebae (11), S. enterica in macrophages (39), and uropathogenic E. coli in superficial bladder epithelial cells (26) are all filamentous. It is, however, pertinent to note that the filamentous cells of S. enterica serovar Typhimurium have distinct FtsZ bands at presumptive midcell locations, and a defect in the histidine biosynthetic pathway is correlated with the observed filamentation phenotype (16). The filamentation phenotype of M. tuberculosis during intracellular growth suggests that the pathogen's cell division process is delayed in response to infection, and this delay could be attributed to compromised function of FtsZTB. Characterization of M. tuberculosis 41 should greatly help us to identify the factors that affect the cell division process during intracellular growth of M. tuberculosis.
This work was supported in part by grants AI48417 (M.R.) and AI41406 (M.V.V.S.M.).
We thank Jaroslaw Dziadek for help with the construction of some plasmids and Zafer Hatahet, William Margolin, Harold P. Erickson, and Marianthi Coronéou for insightful comments and helpful suggestions.