Interaction between FtsW and penicillin-binding protein 3 (PBP3) directs PBP3 to mid-cell, controls cell septation and mediates the formation of a trimeric complex involving FtsZ, FtsW and PBP3 in mycobacteria

Authors


*E-mail joyoti@vsnl.com; Tel. (+91) 332 350 6619; Fax (+91) 332 350 6790.

Summary

In bacteria, biogenesis of cell wall at the division site requires penicillin-binding protein 3 (PBP3) (or FtsI). Using pull-down, bacterial two-hybrid, and peptide-based interaction assays, we provide evidence that FtsW of Mycobacterium tuberculosis (FtsWMTB) interacts with PBP3 through two extracytoplasmic loops. Pro306 in the larger loop and Pro386 in the smaller loop of FtsW are crucial for these interactions. Fluorescence microscopy shows that conditional silencing of ftsW in Mycobacterium smegmatis prevents cell septation and positioning of PBP3 at mid-cell. Pull-down assays and conditional depletion of FtsW in M. smegmatis provide evidence that FtsZ, FtsW and PBP3 of mycobacteria are capable of forming a ternary complex, with FtsW acting as a bridging molecule. Bacterial three-hybrid analysis suggests that in M. tuberculosis, the interaction (unique to mycobacteria) of FtsZ with the cytosolic C-tail of FtsW strengthens the interaction of FtsW with PBP3. ftsW of M. smegmatis could be replaced by ftsW of M. tuberculosis. FtsWMTB could support formation of the FtsZ–FtsW–PBP3 ternary complex in M. smegmatis. Our findings raise the possibility that in the genus Mycobacterium binding of FtsZ to the C-tail of FtsW may modulate its interactions with PBP3, thereby potentially regulating septal peptidoglycan biogenesis.

Introduction

Cell division is a fundamental process central to bacterial propagation. Understanding the mechanistic details of cell division may therefore open hitherto unexplored possibilities of development of new antibacterials. A complex set of proteins likely to form a multiprotein complex is crucial to the formation of the septum during bacterial cell division (Donachie, 1993; Margolin, 2005; Rothfield et al., 2005). FtsZ (Erickson, 1997), a GTP-binding protein, is considered to be the bacterial counterpart of eukaryotic tubulin (de Boer et al., 1992; RayChaudhuri and Park, 1992). It mediates cell division by formation of the Z-ring (Bramhill and Thompson, 1994; Mukherjee and Lutkenhaus, 1994; Erickson et al., 1996). The Z-ring formed by FtsZ serves as a cytoskeletal scaffold for the recruitment of a number of proteins in a sequential manner in Escherichia coli and in a cooperative fashion in Bacillus subtilis (Katis et al., 2000; DiLallo et al., 2003; Errington et al., 2003). In certain instances, as in the case of FtsL, FtsB and FtsQ of E. coli, evidence has been presented in favour of pre-assembly of these proteins before their localization to the septal region (Buddelmeijer and Beckwith, 2004; Aarsman et al., 2005). This complex may also include FtsI and FtsW (Goehring et al., 2005).

In E. coli, FtsA and ZipA bind directly to the C-terminus of FtsZ (Liu et al., 1999; Pichoff and Lutkenhaus, 2002) and are believed to stabilize the Z-rings at mid-cell. However, ZipA and FtsA are not ubiquitous, and counterparts have not been recognized in the globally important pathogen, Mycobacterium tuberculosis, raising the question of how Z-rings may be stabilized in this case. Our previous studies suggest that FtsW is a candidate protein which may be involved in stabilizing Z-rings. FtsZ and FtsW of M. tuberculosis interact directly through oppositely charged residues present in their C-tails (Datta et al., 2002). Rajagopalan and colleagues have recently provided microscopic evidence that in mycobacteria, FtsW colocalizes with FtsZ to the mid-cell (Rajagopalan et al., 2005).

FtsW is a polytopic membrane protein that is present in virtually all bacteria that have a peptidoglycan cell wall (Ikeda et al., 1989; Henriques et al., 1998). It is an essential cell division gene in E. coli (Boyle et al., 1997). In addition to its role in stabilizing the FtsZ ring, FtsW has been suggested to facilitate septal peptidoglycan synthesis by recruitment of penicillin-binding protein 3 (PBP3, or FtsI) to the division site (Weiss et al., 1997; Mercer and Weiss, 2002). PBP3 is the bacterial transpeptidase that is required specifically for cell division (Botta and Park, 1981). Of all the likely components of the divisome, it is the best characterized in terms of its enzymatic activity. It also participates in protein–protein interactions critical for peptidoglycan synthesis (Marrec-Farley et al., 2000; Mercer and Weiss, 2002; Piette et al., 2004). Indirect evidence has suggested that PBP3 and FtsW may interact. However, direct interactions between the two proteins has not yet been demonstrated biochemically. Using a variety of biochemical approaches, we have investigated this possibility using FtsW and the candidate PBP3 of M. tuberculosis, expressed in E. coli. Our studies provide evidence that FtsW interacts directly with PBP3 through two extracytoplasmic loops of FtsW spanning residues 301–320 and 379–386. Pro306 and Pro386 which are conserved across bacterial species, are critical determinants for this interaction. In addition, we demonstrate that FtsZ, FtsW and PBP3 of M. tuberculosis form a ternary complex in vitro, with FtsW participating as a bridging molecule. Bacterial three-hybrid analysis supports the hypothesis that binding of FtsZ to the C-tail of FtsW strengthens the interaction of FtsW with PBP3 likely providing a link between cell division and septal peptidoglycan biosynthesis. These views are strengthened by our observations that ternary interactions among FtsZ, FtsW and PBP3 occur in vivo in Mycobacterium smegmatis and require Pro306 and Pro386 of FtsW. These interactions therefore offer new avenues for exploring strategies for development of chemotherapeutic agents against mycobacteria.

Results

Expression of PBP3 of M. tuberculosis and penicillin-binding activity

The product of the gene Rv2163c present in cosmid MTCY270, one of the collection of clones representing the genome of M. tuberculosis was predicted to represent the counterpart of PBP3 (FtsI) of E. coli in M. tuberculosis, based on sequence similarity with the family of class B high-molecular-mass penicillin-binding proteins (Goffin and Ghuysen, 2002) and the association of the gene with a cluster of genes such as ftsW and ftsZ comprising the division cell wall (dcw) cluster. The ORF Rv2163c designated in the annotated genome of M. tuberculosis shows the presence of a large N-terminal extension (with a likely cytosolic disposition) absent in E. coli PBP3 (Fig. S1). However, we feel that pbp3 could arguably be encompassed by nucleotides 2426805–2425046 of the M. tuberculosis H37Rv genome, based on sequence alignments (Fig. S1) and on the identification of a putative ribosome binding site upstream of the start site, which would give rise to a protein that would be shorter by 93 amino acids than what is predicted now. Definitive proof of the translational site must await further investigation. For the purpose of this report, ‘PBP3’ refers to the 586 (rather than the 679) amino acid PBP3 encoded by nucleotides 2426805–2425046 of the M. tuberculosis H37Rv genome encompassing amino acid residues V94 to T679 (numbering based on the TIGR sequence). The larger 679-amino-acid PBP3 could not be expressed and purified from E. coli. PBP3 and its C-terminal transpeptidase module (Goffin and Ghuysen, 2002) encompassing residues L314 to T679 were expressed in E. coli as 6× His-tagged proteins (Fig. 1A). The protein after purification by nickel-nitrilotriacetic acid (Ni2+-NTA) affinity chromatography bound benzyl-[14C]-penicillin in a concentration-dependent manner (Fig. 1B), indicating that Rv2163c encodes a functional penicillin-binding protein. The C-terminal domain encompassed by residues L314 to T679 bound penicillin with a similar affinity (Fig. 1B), supporting the view that the C-terminal module of PBP3 of M. tuberculosis functions as an independent penicillin-binding entity unlike its E. coli counterpart (Goffin et al., 1996).

Figure 1.

Expression of different constructs of FtsW, PBP3 and analysis of their interactions.
A. Coomassie blue-stained gels of uninduced and induced E. coli cells expressing His-PBP3 (last lane) and His-PBP3 (L314–T679) (middle lane).
B. Fluorograms showing the binding of purified His-PBP3 (upper panel), and His-PBP3 (L314–T679) (lower panel) to different concentrations of benzyl-[14C]-penicillin.
C. The panel on the left shows Amido Black-stained blots of uninduced and induced E. coli membranes coexpressing His-FtsW and myc-PBP3. The panel on the right shows Western blots of membranes from induced cells probed with anti-His or anti-myc antibodies.
D. Cell lysates obtained from E. coli coexpressing His-FtsW and myc-PBP3 (lanes 2, 6, 7) or His-FtsW alone (lane 3) or myc-PBP3 alone (lanes 4, 5) were incubated without (lanes 5, 6) or with anti-His antibody (lanes 2–4) or with an irrelevant antibody (lane 7). Proteins were immunoprecipitated with Protein A/G agarose. Immunoprecipitates were analysed by immunoblotting with anti-myc antibody. The arrowhead indicates the position of myc-PBP3. The first lane is a positive control of expressed myc-PBP3 only. The data shown in B and D are representative of three separate experiments.

Interaction of PBP3 with FtsW

In order to test the possibility that FtsW interacts with PBP3, we expressed the two proteins in tandem in E. coli, both under the control of the T7(lac) promoter, FtsW carrying a hexahistidine tag at its N-terminus, and PBP3 carrying a myc tag at its C-terminus. Both the proteins were expressed under the conditions described (Fig. 1C). Their expression was confirmed by Western blotting using anti-His and anti-myc antibodies. His-FtsW was pulled down from cell lysates using anti-His antibody and protein A/G-agarose. The proteins immobilized along with His-FtsW were separated on SDS-PAGE, electrotransferred and probed with anti-myc antibody. The positive signal obtained after detection by chemiluminescence (Fig. 1D), affirmed our view that PBP3 interacts with FtsW in vivo.

Analysis of the topology of FtsW

FtsW is a polytopic membrane protein. The topologies of the E. coli and Streptococcus pneumoniae FtsW proteins have been determined experimentally (Gerard et al., 2002; Lara and Ayala, 2002). The topology of FtsW of M. tuberculosis was predicted using various available online programs such as tmhmm, sosui and hmmtop. In order to validate the predicted topology, targeted fusions of FtsW were generated with the N-terminal end of the TEM β-lactamase or the chloramphenicol acetyltransferase (CAT) reporters, and susceptibilities of the respective E. coli transformants against ampicillin and/or chloramphenicol were determined (Table S1). E. coli JM105 (lacking chromosomal β-lactamase) had an minimum inhibitory concentration (MIC) of 4 μg ml−1 for ampicillin. The transformants FtsW–Xaa–TEM β-lactamase in which Xaa (the amino acid of FtsW at the fusion junction) was R82, D85, S89, P140, A146, F153, Q209, G224, R234, K301, G311, E320, P379 and P386 respectively, had MICs of 200 μg ml−1 for ampicillin, consistent with the view that these fusion sites were each in the periplasm, because β-lactamase fusion proteins can provide E. coli with ampicillin resistance only if the β-lactamase moiety is translocated to the periplasm (Broome-Smith et al., 1990). E. coli TG1 harbouring FtsW–Xaa–CAT fusions in which Xaa (the amino acid of FtsW at the junction of fusion with CAT) was H58, V110, R120, A179, E190, G255, R257, D270, A281, K282, R342, R346 and R353 respectively, showed chloramphenicol resistance, consistent with the view that these fusion sites were each in the cytoplasm, because CAT-fusion proteins can provide E. coli with chloramphenicol resistance only if CAT is present in the cytosol. Negative validation of the predicted topology was carried out by analysing the β-lactamase constructs generated by fusion to transmembrane (V235, A208 and A300) or cytoplasmic (H58, R257, R342 and R353) amino acids, as well as by analysing CAT constructs fused to transmembrane(Y109, I122, L178, M191, A208, S253, A254, F283, N341 and L354) or periplasmic (Q209, G224, R234 and G311) amino acids. The experimentally determined topology was closest to that predicted by hmmtop, with 10 membrane spanning segments. β-Lactamase fusions to cytoplasmic (H58, R257, R342 and R353) amino acids showed β-lactamase activities in crude extracts of cells comparable to that obtained from E. coli JM105 cells expressing the TEM β-lactamase alone, confirming that the lack of ampicillin resistance was not due to improper folding of fusion proteins. The topology of FtsW derived from these results is presented in Fig. S2.

Mapping of amino acid residues required for interaction of extracytoplasmic loops of FtsW with PBP3

Extracytosolic loops likely harbour the PBP3 binding interface of FtsW. In order to dissect the PBP3-binding elements of FtsW, attempts were made to express several of its extracytosolic loops as His-tagged proteins. These attempts were in several cases unsuccessful. Larger His-tagged constructs were therefore generated in order to test their ability to interact with PBP3. His-FtsW(1-300) was unable to pull down myc-PBP3 (Fig. 2A, lane 2), suggesting that extracytoplasmic loops, if any, encompassed within this domain were not involved in interaction with PBP3. His-FtsW (G255–G524) encompassing the predicted extracytoplasmic loop amino acid residues 301 to 320, and 379 to 386 was able to pull down myc-PBP3 (Fig. 2A, lane 3) as well as its C-terminal module (Fig. 2B) from cell extracts expressing these proteins, suggesting that this domain encompassed a PBP3-interacting region of FtsW. The N-terminal module of PBP3 encompassing amino acid residues V94 to T313 was expressed in E. coli, but did not interact with FtsW (data not shown). PBP3 could not be pulled down in a control tube containing Ni2+-NTA agarose alone (data not shown) indicating that the interaction was specific. Attempts were made to map the amino acid residues within the two loops encompassed by amino acid residues 301 to 320 and 379 to 386 that were likely to be of importance in interaction with PBP3. When compared with E. coli FtsW, the region encompassed by the amino acid residues 301 to 320 showed stretches of conserved amino acid residues (Fig. S3). Derivatives of His-FtsW (G255–G524) lacking residues F312 to F314 or G318 to E320 were capable of interacting with PBP3, whereas interaction was abrogated by a deletion encompassing residues Y304 to P306 (data not shown), suggesting that amino acid(s) within the region encompassing Y304 to P306 were necessary for interaction with PBP3. Each of the three residues within this region was then mutated successively to alanine. Only the P306A mutation resulted in a loss of binding of His-FtsW (G255–G524) to myc-PBP3 (Fig. 2A, lane 6). Comparison of the sequences of FtsW proteins available in the database showed that the putative periplasmic loop spanning amino acid residues P379 to P386 contains several conserved amino acid residues including P379 and P386. Proline residues often occur near protein–protein interaction sites (Kini and Evans, 1995). His-FtsW (L361–G524) could bind myc-PBP3 (Fig. 2A, lane 7). Mutation of P379 to Ala did not abrogate the interaction (data not shown). However, mutation of P386 to Ala led to an abrogation of interaction (Fig. 2A, lane 8), suggesting that it is one of the residues involved in interactions with PBP3. In each case, expression of the His-FtsW constructs being studied was verified by reprobing of blots with anti-His antibody (Fig. 2C). His-FtsW (G255–G524) bearing the double mutation P306A, P386A, did not bind myc-PBP3 (Fig. 2A, lane 4), confirming the importance of these two proline residues in the interaction.

Figure 2.

Interaction of different domains of FtsW with PBP3. Lysates of E. coli expressing His-tagged proteins derived from domains of FtsW encompassing the indicated amino acid residues (wt, or mutants) were incubated with Ni2+-NTA-agarose beads in order to immobilize His-FtsW proteins. Lysates of E. coli expressing myc-PBP3 (A) of myc-PBP3 (L314–T679) (B) were incubated with immobilized His-FtsW constructs and the precipitates containing proteins bound to the agarose beads were analysed by immunoblotting using anti-myc antibody. The first lanes of A and B represent positive controls with purified myc-PBP3 and -PBP3 (L314–T679) respectively. The blot in A was reprobed with anti-His antibody to ensure equal loading of all the His-FtsW proteins (C). (D) Incubations of Ni2+-NTA-agarose-bound His-FtsW (G255–G524) with cell lysates expressing myc-PBP3 were carried out in the absence or presence of peptide J4 or a scrambled (scr.) peptide (at the indicated molar ratios). The precipitates containing proteins bound to the beads were analysed by immunoblotting using anti-myc antibody. The first lane of D represents a negative control containing only Ni2+-NTA-agarose-bound His-FtsW (G255–G524). (E) His-PBP3 or an irrelevant His-tagged protein (encoded by the ORF Rv0129c) (of similar pI) was allowed to bind biotinylated peptide J4 as described under Experimental procedures. Peptide-bound PBP3 was detected by pull-down with streptavidin–agarose followed by Western blotting with anti-His antibody. The data shown in A–E are representative of three separate experiments.

The putative extracytoplasmic loop encompassed by residues 376–386 was too short to be tested for binding to PBP3 as an epitope-tagged protein. The role of this loop was therefore confirmed by testing the ability of the peptide J4 encompassing residues 376–386 of FtsW to inhibit the binding of FtsW (G255–G524) to myc-PBP3. J4 could inhibit the interaction in a dose-dependent manner (Fig. 2D). No inhibition was obtained in the control tube containing a scrambled sequence. The presence of peptide J4 did not affect the penicillin-binding ability of PBP3 (data not shown). Alternatively, the biotinylated peptide J4 was incubated with His-PBP3 for 6 h at 25°C and the association of the peptide with PBP3 was visualized by pulling down the complex with streptavidin–agarose, followed by electrophoretic separation and detection with anti-His antibody (Fig. 2E). Peptide J4 was found to interact with PBP3. A variety of in vitro methods therefore supported the view that the region encompassed by residues 376–386 of protein FtsW represents a site of interaction of FtsW with PBP3. The residues 376–386 of FtsW therefore appeared likely to harbour determinants crucial to the FtsW-PBP3 binding interface.

Interactions between FtsZ, FtsW and PBP3

Given the fact that FtsW interacts with both FtsZ (Datta et al., 2002) and PBP3, we tested the hypothesis that these proteins are capable of forming a ternary complex. We expressed S-tagged FtsZ, myc-tagged PBP3 and His-tagged FtsW (G255–G524). The expression of full-length FtsW was poor and FtsW was susceptible to proteolytic degradation (Datta et al., 2002). We therefore expressed FtsW (G255–G524) which harbours both the FtsZ- and the PBP3-interacting determinants. Cells expressing the individual constructs were mixed prior to disruption, followed by immunoprecipitation of each of these proteins separately. The method was similar to the strategy used by Noirclerc-Savoye et al. (2005) for studying the interactions between DivIB, DivIC and FtsL of S. pneumoniae. The identity of the co-immunoprecipitated proteins was verified by Western blotting. Immunoprecipitation with anti-S-tag agarose, followed by immunoblotting with anti-His [for FtsW (G255–G524)] (Fig. 3, blot 1) or anti-myc (for PBP3) (Fig. 3, blot 3) showed that both FtsW and PBP3 were part of a complex with FtsZ. When cells expressing FtsW (G255–G524) were omitted from the mixture prior to cell disruption, FtsZ could not pull down PBP3 from the lysates (blot 4). FtsW therefore appeared to be the bridging partner in these interactions resulting in a trimeric complex between the three proteins. Our earlier observations have suggested that a stretch of aspartate residues at the C-terminal end of FtsZ are crucial for its interactions with FtsW of M. tuberculosis (Datta et al., 2002). In harmony with this, the FtsZ (ΔD367–D370) mutant was unable to co-immunoprecipitate FtsW (blot 1, last lane) or PBP3 (blot 3, last lane). In a similar manner, precipitation with Ni2+-NTA agarose (Fig. 3, blots 5–7) followed by immunoblotting with anti-myc (blot 5) or anti S-tag (blot 6) antibodies confirmed the formation of a trimeric complex between FtsW (G255–G524), FtsZ and PBP3. The interaction with PBP3 diminished upon mutation of either P306 to A or P386 to A (blot 5). The ability of the FtsW construct to co-immunoprecipitate PBP3 was inhibited almost completely in the case of the P306A, P386A double mutant (blot 5).

Figure 3.

Interactions among FtsZ, FtsW and PBP3. E. coli cells overexpressing individually, His-tagged FtsW (G255–G524), myc-tagged PBP3 and S-tagged wild-type FtsZ (WT) [or its mutant (ΔD367–D370)] were mixed, lysed with Cell Lytic™ BII and solubilized cell lysate was precipitated with anti-S-tag agarose (blots 1–3), followed by Western analysis with the indicated antibodies. The fourth blot represents a similar set of experiments, except that FtsW was omitted. Western analysis with anti-S tag antibody (second blot) confirmed equal loading of FtsZ and its mutant. In a separate set of experiments, E. coli cells overexpressing individually, myc-tagged PBP3, S-tagged wild-type FtsZ (WT) and His-tagged FtsW [G255–G524 (WT)] (or its mutants as indicated in the figure), were mixed, lysed with Cell Lytic™ BII and solubilized cell lysate was precipitated with Ni2+-NTA agarose (blots 5–7). The agarose-bound proteins were immunoblotted with anti-myc or anti-S tag or anti-His antibody to detect PBP3 or FtsZ or FtsW respectively. Western analysis with anti-His antibody (blot 7) confirmed equal loading of FtsW and its mutants. E. coli cells expressing individual wild-type proteins (positive controls) were lysed and immunoblotted with the above-mentioned antibodies separately in order to identify the position of PBP3, FtsZ and FtsW (indicated by arrows). Data shown are representative of results obtained in three separate experiments.

Bacterial two-hybrid analysis

The bacterial adenylate cyclase two-hybrid analysis (BACTH) based on the method of Karimova et al. (2005) was used to further analyse interactions between FtsW, FtsZ and PBP3. Pairwise interactions between FtsZ and FtsW, as well as between FtsW and PBP3 were analysed by co-transforming the E. coli cya strain DHM1 with pairs of recombinant plasmids expressing the T25 and T18 hybrids. The efficiencies of functional complementation between the different hybrids were determined by β-galactosidase assays. FtsZ could interact with FtsW (Fig. 4A and B). This interaction was abrogated when residues D367–D370 were deleted from FtsZ (Fig. 4B) or when residues R510–R514 were deleted from the FtsW construct (Fig. 4A), confirming the importance of these stretches of oppositely charged residues in the interaction. PBP3 could also interact with FtsW (Fig. 4C). However, this interaction was weakened twofold when either P306 or P386 of FtsW was individually mutated. When both the residues were mutated, β-galactosidase activity was reduced to the level of the negative control (Fig. 4C) suggesting the involvement of both these proline residues of FtsW in its interaction with PBP3. In order to study the possible ternary interaction between FtsZ, FtsW and PBP3, FtsZ (or its mutant) was coexpressed with FtsW in the vector pKT25. FtsW, when coexpressed with FtsZ, showed a higher efficiency of interaction with PBP3 expressed in pUT18C than FtsW alone with PBP3 (Fig. 4D). However, FtsZ alone (in the absence of FtsW) did not interact with PBP3 (last bar in Fig. 4D). The specificity of the effect of FtsZ in enhancing interaction between FtsW and PBP3 was borne out by the fact that the FtsZ (ΔD367–D370) mutant (which lacks the ability to interact with FtsW) was not able to enhance interaction when coexpressed with FtsW (Fig. 4D). The knowledge that particular deletions (or point mutations) selectively abolished interactions between specific partners, provide insights into critical determinants of intrinsic associations between these molecules.

Figure 4.

Bacterial three- or two-hybrid analysis to study the interactions between FtsW, FtsZ and PBP3. E. coli DHM1 cells were co-transformed with different constructs (or control vectors) as detailed below.
A. pUT18C (vector, negative control) or FtsW in pUT18C or its mutant [FtsW (ΔR510–R514)] in pUT18C were co-transformed with FtsZ in pKNT25.
B. pKNT25 (vector, negative control) or FtsZ in pKNT25 or FtsZ (ΔD367–D370) in pKNT25 was co-transformed with FtsW in pUT18C.
C. PBP3 in pUT18C was co-transformed with pKT25 (vector, negative control) or FtsW (or its mutants as indicated) in pKT25.
D. PBP3 in pUT18C was co-transformed with vector only (negative control) or FtsW, or FtsZ or FtsW along with FtsZ [WT or (ΔD367–D370)] in pKT25 in order to study the interaction between FtsZ and PBP3. In each case, functional complementation between the hybrid proteins was quantified by measuring the β-galactosidase activities in toluene-treated E. coli cells harbouring the plasmids as indicated in the figure. Each bar represents the mean value ± SD from three independent experiments.

Prolines 306 and 386 of FtsW of M. tuberculosis are necessary for viability

The preceding studies have focused primarily on in vitro associations of FtsW, FtsZ and PBP3. We wanted to test the relevance of these interactions in vivo. A pairwise blast analysis (Altschul et al., 1997) showed that M. smegmatis FtsZ, FtsW and PBP3 were 91%, 66% and 78% identical; and 95%, 75% and 89% similar in amino acid sequence to their M. tuberculosis counterparts (Figs S4–S6). The hydrophilic C-tails of FtsW and FtsZ shown to mediate interaction between the two proteins of M. tuberculosis were conserved in their M. smegmatis counterparts. The two extracytoplasmic loops of FtsW predicted to mediate interaction with PBP3 were also conserved in the M. tuberculosis and M. smegmatis proteins. Taking these into consideration we chose the rapidly growing mycobacterium, M. smegmatis as the model system to test whether the associations are of relevance in vivo in mycobacteria.

In order to test whether FtsW of M. tuberculosis can substitute for FtsW of M. smegmatis, and to identify amino acid residues critical for interaction of FtsW with PBP3, we used the two-step homologous recombination protocol described by Rajagopalan et al. (2005) and Chauhan et al. (2006). Merodiploid single crossover (SCO) strains carrying integrated copies of N-terminal, His-tagged wild-type ftsW of M. tuberculosis (ftsWMTB) or ftsWMTB genes encoding the P306A, or P386A or P306A, P386A mutants, were generated. The His-tag was incorporated in order to facilitate detection as well as immunoprecipitation of FtsWMTB. To test the consequences of mutating P306 and/or P386, we attempted to delete ftsW of M. smegmatis in the presence of an integrated copy of either wild-type or mutant ftsWMTB. Double crossovers (DCOs) were selected and analysed by PCR using primers specific for either the integrated copy of M. tuberculosis ftsW or the chromosomal copy of M. smegmatis ftsW. DCOs carrying chromosomal M. smegmatis ftsW deletions were obtained in cells expressing the integrated copy of either wild-type ftsWMTB or ftsWMTB encoding the P306A or the P386A mutations. By contrast only wild-type M. smegmatis ftsW patterns were obtained in 30 DCOs carrying ftsWMTB P306A, P386A (ftsW *). Failure to delete the chromosomal ftsWSMEG in M. smegmatis carrying an integrated copy of ftsW * supported our contention that both P306 and P386 are necessary for FtsW to fulfil its in vivo function. Mutation of either one of the two proline residues was likely less deleterious for the cells than deletion of both the proline residues. DCOs carrying ftsWMTB(WT), or ftsWMTB(P306A) or ftsWMTB(P386A) were grown in Middlebrook 7H9 broth. Growth was reduced twofold in strains carrying ftsWMTB(P306A) or ftsWMTB(P386A) compared with the strain harbouring ftsWMTB (Fig. S7).

FtsW interacts with FtsZ and PBP3 in M. smegmatis

Double crossovers were used to analyse the interactions between FtsZ, FtsW and PBP3 in vivo. Expression of His-FtsWMTB did not cause any dramatic change in morphology, and the viability of the strain was comparable to that of the wild-type strain (data not shown), indicating that the production of His-FtsWMTB in a strain lacking the chromosomal copy of M. smegmatis FtsW was not toxic to the cells. Lysates from cells expressing His-FtsWMTB (or its mutants) were immunoprecipitated with anti-His antibody. Western analysis with anti-PBP3 antibody showed that the P306A or the P386A mutants of FtsWMTB were less efficient in pulling down PBP3 than the wild-type FtsWMTB (Fig. 5A). This supported the view that both P306 and P386 of FtsW play a role in defining the PBP3-interacting interface of FtsW. Reprobing with anti-His antibody confirmed that the expression of His-FtsW was comparable in each case (Fig. 5A, lower blot). In the reverse experiment, lysates were immunoprecipitated with anti-PBP3 antibody and probed with anti-His antibody. Western analysis again confirmed that the P306A and the P386A mutants of FtsW were less proficient in interacting with PBP3 than the wild type (Fig. 5B). The specificity of the band visualized was confirmed by the fact that immunoprecipitation of lysates from wild-type M. smegmatis mc2 155 with anti-PBP3 antibody followed by Western analysis with anti-His antibody did not give any band (Fig. 5C). The role of P306 and P386 in regulating ternary interactions between FtsZ, FtsW and PBP3 was then evaluated using anti-PBP3 and anti-FtsZ antibodies which did not cross-react (data not shown). Immunoprecipitation with anti-PBP3 followed by Western analysis with anti-FtsZ (Fig. 5D) and vice versa (Fig. 5E) showed weakened FtsZ–PBP3 interaction in DCOs carrying ftsWMTB encoding the P306A or P386A mutants compared with the wild type. The expression of PBP3 and FtsZ in cell lysates was comparable in all the DCOs (lower blots of panels D and E respectively).

Figure 5.

Interaction of FtsZ and PBP3 with FtsW in M. smegmatis. M. smegmatis DCO strains (A, B, D and E, bearing His-tagged wild-type or mutated ftsW of M. tuberculosis as indicated in the figure) were grown, lysed and membranes were solubilized with Triton X-100 as described under Experimental procedures. Solubilized membranes were immunoprecipitated with anti-His (A) or anti-PBP3 antibody (B), followed by immunoblotting with anti-PBP3 (A) or anti-His (B) antibody. The lower blot represents reprobing with anti-His (A) or anti-PBP3 (B) antibody. The first lane of panel A represents immunoprecipitation from the solubilized membranes of DCO (wild type) with anti-His antibody followed by immunoblotting with pre-immune sera. The last lane of panel B represents immunoprecipitation from the solubilized membranes of DCO (wild type) with pre-immune sera followed by immunoblotting with anti-His antibody. Panel C (negative control) represents immunoprecipitation from the solubilized membranes of untransformed M. smegmatis mc2 155 only. For D and E, immunoprecipitation was carried out with anti-PBP3 or anti-FtsZ antibody followed by immunoblotting with anti-FtsZ or anti-PBP3 respectively. The first lane of each blot represents immunoprecipitation with pre-immune sera only. The lower blots of D and E represent reprobing with anti-PBP3 or anti-FtsZ respectively.

Depletion of FtsW of M. smegmatis by antisensing

In order to analyse whether FtsW acts as a bridging molecule within a ternary complex involving FtsZ, FtsW and PBP3, in vivo, it was necessary to develop an FtsW-depleted system. Construction of an ftsW knockout was not possible because ftsW is an essential gene. An antisense construct of FtsW in pMIND, pJB223, was designed. pJB223 carried the whole of the ftsW gene of M. smegmatis in antisense orientation under the control of the tetRO region. Conditional depletion of ftsW from M. smegmatis harbouring pJB223 was achieved by the addition of tetracycline (Blokpoel et al., 2005). Growth in the presence of tetracycline was followed up to 36 h. Cell lysis was recorded by a decrease in absorbance at 600 nm beginning 24 h after growth (data not shown). No lysis was observed at this time point when cells were grown in the absence of tetracycline. We chose to record the consequences of FtsW depletion 16 h after growth in the presence of tetracycline. At this time point, Northern analysis (as described by Ji et al., 1999) using both sense and antisense probes designed on the ftsW sequence, confirmed the presence of antisense RNA and the absence of sense RNA (data not shown). Cells were elongated, but no lysis had occurred. No change in morphology was observed in control cultures of M. smegmatis (without the inducible construct) before and after the addition of tetracycline (data not shown), confirming that the altered morphology was not a consequence of exposure to tetracycline.

M. smegmatis mc2 155 harbouring pJB223 was grown in the absence or in the presence of tetracycline. Wheat germ agglutinin (WGA)-Alexa 488 staining was done in order to visualize cell wall and septa. WGA binds specifically to N-acetylglucosamine in the outer peptidoglycan layer for Gram-positive bacteria (Sizemore et al., 1990) including M. smegmatis (Dasgupta et al., 2006). Tetracycline-induced FtsW depletion was associated with impaired septum formation inferred from the inability to detect WGA-Alexa 488 staining between nucleoids (Fig. 6). 85 ± 5% (n = 250) of the cells after ftsW silencing contained more than one nucleoid per cell as observed by DAPI staining. On an average, each cell contained 2.8 nucleoids after FtsW depletion.

Figure 6.

Fluorescence microscopy of M. smegmatis mc2 155 before and after depletion of FtsW. M. smegmatis mc2 155 was transformed with pMIND carrying ftsW. Cells were grown without (–TET) or with (+TET) tetracycline. Upper micrographs: DAPI staining; middle micrograph: WGA-Alexa 488 (green) staining; lower micrographs: overlay of DAPI (blue) and WGA-Alexa 488 (green) staining. Each bar represents 2 μm. The cell in the inset in each panel is shown after magnification.

Localization of FtsZ and PBP3 in FtsW-depleted M. smegmatis

In order to study the septation defect in more detail, immunofluorescence microscopy was performed to localize FtsZ and PBP3. FtsZ localized at mid-cell in uninduced cells (Fig. 7A). When FtsW was depleted (as described above), FtsZ was still observed between nucleoids (Fig. 7B). However, cell septation was not complete (Fig. 7B). In the wild-type or –tet cells, one FtsZ ring per cell was observed. When FtsW was depleted (+tet), we observed one to six rings per cell, with the majority of the filaments containing three rings per filament. In uninduced cells, PBP3 localized to mid-cell as well as at the poles (Fig. 8). Staining at the poles was visible in small cells, suggesting that the new cell pole had likely retained some PBP3 from the previous division. This finding was similar to that observed in the case of E. coli PBP3 (Weiss et al., 1997). When FtsW was depleted from M. smegmatis/pJB223 by the addition of tetracycline, PBP3 could no longer be visualized by immunofluorescence microscopy (data not shown). DAPI staining (Fig. 7B, top panel) verified that the filaments exhibited proper nucleoid segregation even though PBP3 failed to localize, indicating that the cells were healthy. Control experiments using pre-immune sera of FtsZ or PBP3 did not show any staining of cells (data not shown).

Figure 7.

Visualization of FtsZ in M. smegmatis by fluorescence microscopy. M. smegmatis mc2 155 was transformed with pMIND carrying ftsW followed by induction with tetracycline in order to inactivate ftsW conditionally. Transformed cells before (A) or after (B) induction were stained with DAPI (blue, left panel) for visualizing nucleoids. Immunolocalization of FtsZ was carried out by incubation with anti-FtsZ antibody as described under Experimental procedures followed by staining with Alexa 488-conjugated rabbit IgG (green, middle panel). The panel on the right is a merge of the first two panels showing the presence of FtsZ between nucleoids. The failure of cells to separate after silencing ftsW is visible (B, merged image). Each bar represents 2 μm. The cell in the inset is shown after magnification.

Figure 8.

Visualization of PBP3 in M. smegmatis by fluorescence microscopy. M. smegmatis mc2 155 was transformed with pMIND carrying ftsW. Transformed cells before induction with tetracycline were stained with DAPI (top panel) for visualizing nucleoids or Alexa 488 (green, middle panel). Immunolocalization of PBP3 was carried out by incubation with anti-PBP3 antibody as described under Experimental procedures followed by staining with Alexa 488-conjugated rabbit IgG (green, middle panel). The bottom panel is a merge of the first two panels showing the presence of PBP3 at the division site between nucleoids. PBP3 could not be visualized by immunostaining after induction with tetracycline to silence ftsW. No panel has therefore been provided for cells after induction with tetracycline. Each bar represents 2 μm. The white and orange arrows indicate the positioning of PBP3 at poles and mid-cell respectively.

Complex formation involving FtsZ and PBP3 in FtsW-depleted M. smegmatis

We designed experiments to test the status of the complex between FtsW, PBP3 and FtsZ in FtsW-depleted M. smegmatis. In the presence of tetracycline, PBP3 could pull down FtsZ in cells carrying the control vector (pMIND) only (Fig. 9A, lane 3) but not in cells from which FtsW had been depleted by antisensing (pMIND-FtsW) (Fig. 9A, lane 2). Conversely, anti-FtsZ antibody could not pull down PBP3 in FtsW-depleted cells (Fig. 9B, lane 2). These results clearly suggested that the in vivo complex formation between FtsZ and PBP3 was specific. Immunoblotting of the cell lysate obtained after FtsW depletion did not show any alteration in the content of either FtsZ or PBP3 (Fig. S8) indicating that the failure to detect interaction between FtsZ and PBP3 in the absence of FtsW was due to the absence of neither FtsZ nor PBP3.

Figure 9.

Complex formation between FtsZ, FtsW and PBP3 in M. smegmatis. M. smegmatis mc2 155 was transformed with pMIND alone (lane 3 of A and B) or pMIND carrying ftsW (lanes 1, 2 and 5 of A and B). Transformed cells were left uninduced (–), or induced (+) with tetracycline to induce conditional inactivation of ftsW. Membranes were prepared from M. smegmatis (lane 4 of A and B) as well as from transformed cells before and after induction with tetracycline. Solubilized membranes were immunoprecipitated with anti-PBP3 (lanes 1–4 of A) or anti-FtsZ (lanes 1–4 of B). In control experiments, solubilized membranes obtained from transformed cells before induction (–) with tetracycline were immunoprecipitated with pre-immune sera of PBP3 (lane 5, A) or FtsZ (lane 5, B). Immunoprecipitates were blotted with anti-FtsZ (A) or anti-PBP3 (B) antibody. Lane 6 of A and B represents Western analysis of recombinant, purified FtsZ and PBP3 with their respective antibodies.

Discussion

Cell division and the formation of the wall peptidoglycan are steps which are critical to the propagation and survival of most bacterial species. A multicomponent protein machinery has been postulated to assemble in an ordered fashion in E. coli to co-ordinate cell septation at mid-cell. The GTP-binding tubulin-like protein FtsZ serves as a cytoskeletal scaffold for the recruitment of a battery of proteins to the division site. FtsW is believed to stabilize the Z ring at the membrane. Our own studies have shown that M. tuberculosis FtsW interacts directly with FtsZ through strings of oppositely charged residues at their C-termini (Datta et al., 2002). In this study, we have analysed the interaction between FtsW and its likely binding partner PBP3 on the extracytoplasmic side of the membrane. We also designed experiments to test the hypothesis that the polytopic membrane protein FtsW is a bridging molecule linking cytokinesis with septal cell wall biosynthesis.

Using the algorithm hmmtop (Tusnady and Simon, 1998; 2001), FtsW was predicted to span the membrane 10 times with both N- and C-termini in the cytoplasm. The predicted topology was confirmed by generating a series of β-lactamase and CAT fusions. Our results did not support a model predicting a large extracytosolic loop between transmembrane segments 7 and 8 proposed in the case of FtsW of E. coli (Lara and Ayala, 2002) and S. pneumoniae (Gerard et al., 2002). In vitro pull-down, competition and peptide binding assays suggested that the two loops spanning residues 301 to 320 and 379 to 386 interact directly with PBP3 with the conserved residues P306 and P386 being critical for the interaction with PBP3. Pastoret et al. (2004) have shown that simultaneous mutation of both P368 and P375 of E. coli FtsW (the counterparts of P379 and P386 of M. tuberculosis FtsW respectively) prevents the recruitment of PBP3, suggesting that this short extracytoplasmic loop is likely to be important for FtsW–PBP3 interactions in vivo in E. coli as well. However, in mycobacteria, the larger loop carrying P306 also appeared likely to be involved in FtsW–PBP3 interaction, whereas no reports suggest that this loop is necessary for FtsW to fulfil its function in E. coli. Our views were further strengthened by the following observations. ftsWMTB could replace ftsWSMEG without retardation of growth or loss of viability. DCOs could still be obtained when the chromosomal copy of ftsWSMEG was inactivated in the presence of an integrated copy of ftsWMTB encoding point mutations either at P306 or at P386, whereas ftsWMTB encoding the P306A, P386A double mutant could not replace ftsWSMEG.

Earlier work conducted in several laboratories, mostly with E. coli, has relied on mutating a particular component of the divisome and analysing the ability of the mutant to recruit other components, as the preferred approach to studying interactions between proteins of the multicomponent cell division machinery. Direct biochemical evidence of interaction between divisome-associated proteins has been shown only recently in the case of FtsQ, FtsB and FtsL in E. coli (Buddelmeijer and Beckwith, 2004), and more recently between DivIC, DivIB and FtsL of S. pneumoniae (Noirclerc-Savoye et al., 2005). In spite of its limitations, in vitro analyses of protein–protein interactions offer a method of validating whether interacting partners are by themselves sufficient for an interaction to occur, or whether additional players are required. Our detailed in vitro studies have provided insight into how direct interactions between FtsZ and FtsW; and FtsW and PBP3 occur. Our studies are the first to demonstrate biochemically, direct protein–protein interactions between FtsW and PBP3. We observed that in in vitro assays the C-terminal module of PBP3 interacts with FtsW, whereas the N-terminal module encompassing residues V94 to T313 (harbouring several amino acid residues conserved in E. coli PBP3 and predicted to be important for localization of E. coli PBP3 to mid-cell) did not interact with FtsW. However, as we could not express any construct harbouring the first 93 amino acids of PBP3, a definitive answer regarding the role of the N-terminal 93 amino acids in mediating interaction of PBP3 with FtsW must await further experimentation.

In order to establish the physiological significance of the interaction between FtsW and PBP3 in mycobacteria, we used M. smegmatis as a model system. The relevance of the FtsZ–FtsW and the FtsW–PBP3 interaction was established by silencing ftsW in M. smegmatis. Under these conditions, PBP3 failed to localize to mid-cell. Although FtsZ could still be observed at mid-cell, cell septation was compromised. Localization of FtsZ at mid-cell was therefore not sufficient for cell division.

We propose that in mycobacteria, FtsZ, FtsW and PBP3 form a complex. The following lines of evidence support this view. (i) In vitro reconstitution experiments showed that FtsW, FtsZ and PBP3 form a ternary complex. FtsZ and PBP3 were not capable of interacting in the absence of FtsW, suggesting that FtsW is a bridging molecule. (ii) Bacterial two/three-hybrid analyses suggested the likely existence of FtsZ–FtsW–PBP3 interactions. FtsW could interact with PBP3 as determined by β-galactosidase assays. Coexpression of FtsZ with FtsW enhanced β-galactosidase activity, suggesting strengthening of the FtsW-PBP3 interactions by the presence of FtsZ. (iii) Expression of a chromosomal copy of His-tagged ftsWMTB in M. smegmatis (inactivated in its chromosomal ftsW gene), followed by immunoprecipitation with anti-His antibody and Western blotting with anti-FtsZ and anti-PBP3 antibody, confirmed that the FtsW immunoprecipitate contained both FtsZ and PBP3. The fact that anti-PBP3 could pull down FtsZ and vice versa from extracts of these cells strengthened the view that ternary interactions among FtsZ, FtsW and PBP3 occur in vivo. The weakening of the band intensities in cells carrying ftsWMTB(P306A) or ftsWMTB(P386A), suggested that both these proline residues of FtsW were important for ternary complex formation. (iv) ftsW was silenced in M. smegmatis using an antisensing approach. In control cells (where ftsW had not been silenced), FtsZ could pull down PBP3 and vice versa, whereas the results of pull-down experiments were negative when ftsW was silenced. These results confirmed the likely existence of an FtsZ–FtsW–PBP3 ternary complex in mycobacteria, and suggested that the interactions are of physiological relevance.

The present findings emphasize how little we understand of cell division in mycobacteria. In recent years the formation of protein complexes involving several divisome proteins has been demonstrated in E. coli (Goehring et al., 2005). FtsQ, FtsL and FtsB form complexes (Buddelmeijer and Beckwith, 2004), and a subassembly of FtsA, PBP3 and FtsN has also been suggested (Corbin et al., 2004). In mycobacteria, it is possible that some of the divisome proteins may exist in the cell as multiprotein complexes. We observe that FtsZ is recruited to mid-cell even when FtsW is depleted. PBP3, on the other hand, localizes to mid-cell only in the presence of FtsW. The present study suggests that binding of FtsZ to FtsW could potentially modulate the association of FtsW with PBP3. While initial targeting of FtsZ to the predivisional site is probably FtsW-independent, the interaction between FtsW and FtsZ occurring through their C-tails (unique to mycobacteria) is of likely physiological relevance. The report by Rajagopalan et al. (2005) that ftsZ * encoding FtsZ with mutated C-terminal residues needed for interaction with FtsW did not complement, supports this view. This suggests that a complex network of interactions between the components of the divisome likely co-ordinates cell division with biogenesis of septal peptidoglycan. The current hypothesis is that PBPs localize at the division site mainly through multienzyme complexes. Cytoskeletal proteins have been speculated to be part of such complexes (Cabeen and Jacobs-Wagner, 2005). RodA and FtsW have been projected as candidate bridging molecules in the two complexes involved in cell elongation and cell division respectively. Evidence of the existence of the FtsZ–FtsW–PBP3 ternary complex in mycobacteria makes it tempting to speculate that the formation of this complex imparts FtsZ with the ability to regulate the activity of PBP3 (which in E. coli, is to modify the existing peptidoglycan to an inert polar peptidoglycan). Rigorous experimentation is necessary to elucidate the precise function of PBP3 in peptidoglycan remodelling in mycobacteria; and to elucidate how peptidoglycan is generated. The present findings encourage further investigation into how the components of the cell division machinery assemble in mycobacteria and how multiprotein complexes regulate septal peptidoglycan biosynthesis.

Experimental procedures

Molecular biological procedures

Standard procedures for cloning and analysis of DNA, PCR, electroporation and transformation were used (Sambrook et al., 1989). Enzymes used to manipulate DNA were from Roche Applied Science, Mannheim, Germany. All constructs made by PCR were sequenced to verify their integrity. E. coli strains were routinely grown in Luria broth (LB). All constructs were verified by sequencing.

The pbp3 (Rv2163 c) gene of M. tuberculosis (encoding amino acids V94 to T679, TIGR numbering) was amplified from cosmid MTCY270 and cloned in pET28a (Novagen) using asymmetric NcoI and XhoI sites or in pBAD-Myc/HisB (Invitrogen) (between BglII and EcoRI) to generate plasmids pJB301 and pJB302 respectively. A construct for expression of PBP3 (L314–T679) was generated by cloning an appropriate shorter amplicon between the NdeI and EcoRI sites of pET28a to generate plasmid pJB303. In order to obtain myc-tagged PBP3 (L314–T679) under the control of the T7 promoter, the gene was amplified using pJB301 as template and the product was cloned between the asymmetric BglII and EcoRI sites in pBAD-Myc/HisB to generate plasmid pJB304. Next, the gene was amplified using pJB304 as template and cloned between the NdeI and EcoRV sites of pET29a(+) to generate plasmid pJB305 for expression of myc-tagged PBP3 (L314–T679). A construct for dual expression of FtsW and PBP3 was generated in the vector pET-DUET (Novagen). The 1.6 kb NcoI-EcoRI fragment encoding the ftsW gene was excised from pJB201 (Datta et al., 2002) and cloned in pET-DUET to give plasmid pJB306. The pbp3 gene was amplified from plasmid pJB302 and cloned between the BglII and EcoRV sites of the plasmid pJB306 to generate plasmid pJB307. The primers are given in Table S2.

The constructs for expression of His-tagged proteins of different domains of FtsW were generated using the primer pairs depicted in Table S3 and pJB201 as template. The amplified PCR products were cloned in pET28a.

Mutants within the derivatives of FtsW were generated by overlap extension PCR. The primers used are depicted in Table S4. The construct for expression of FtsZ bearing a C-terminal S-tag was generated by amplification of the ftsZ gene from plasmid pJB101 (Datta et al., 2002) using the primer pair 5′-TATGGATCCATATGACCCCCCCGCACAACTA-3′ (sense) and 5′-TTTGGTACCGCGGCGCATGAAG-3′ (antisense), and cloning between the NdeI and KpnI sites (in bold) of the vector pACYC-DUET (Novagen) to generate plasmid pJB102. The mutant of FtsZ deleted in residues D367–D370 was generated by overlap extension PCR as described earlier (Datta et al., 2002) and cloned into pACYC-DUET as described above to generate pJB103.

Construction of β-lactamase fusions with truncated FtsW derivatives

PCR was performed using the sense primer 5′-ATCGGATCCATATGCTAACCCGGTTGCTGC-3′, antisense primers carrying portions of the ftsW gene (see Table 5) and pJB201 as template. The amplified products were cloned between the BamHI (in bold) and PvuII sites of pJBS633 which carries the mature TEM β-lactamase encoding blaM (Broome-Smith and Spratt, 1986). E. coli JM105 was transformed with the ligation mixture, and transformants growing on LB agar plates containing 50 μg ml−1 kanamycin were chosen for further analysis. Transformants containing in-frame FtsW-β-lactamase fusions were detected by their ability to grow when patched with toothpicks on to agar containing 200 μg ml−1 ampicillin (Broome-Smith and Spratt, 1986). The nucleotide sequences across the β-lactamase fusion junctions were determined. The MICs of ampicillin for E. coli JM105 containing FtsW–Xaa–TEM β-lactamase fusions were determined by spotting 4 μl of a 1:105 dilution of an overnight culture (approximately 40 bacteria) on LB agar plates containing a range of doubling concentrations of ampicillin (Broome-Smith and Spratt, 1986). The MIC was the lowest concentration of ampicillin that prevented the growth of bacterial colonies.

Construction and analysis of targeted CAT fusions

The cat gene was PCR-amplified from pACYC184 using the forward primer 5′-AGGGTACCAAAAAAATCACTGGATATA-3′ (KpnI site in bold) and reverse primer 5′-ATAAAGCTTCGCCCCGCCCTGCCACTC-3′ (Hind III site in bold) and inserted between the KpnI and HindIII sites of pBADMycHisA generating pBAD-CAT. CAT constructions were created by targeted PCR fusion with derivatives of FtsW, using the forward primer 5′-ATAGATCTGTGCTAACCCGGTTGCTG-3′ containing a BglII site (in bold) and reverse primers encoding portions of the ftsW gene (Table S6). The constructs were transformed in E. coli TG1 containing 100 μg ml−1 ampicillin and tested for resistance to chloramphenicol as follows. An overnight culture in LB medium was diluted 10-fold with fresh medium and allowed to grow for 3 h. Induction was then carried out by the addition of 0.05% l-arabinose for 1 h, and 4 μl of a 104 dilution of the cultures was spotted on plates containing different concentrations of chloramphenicol in combination with 0.2% l-arabinose. Growth was observed after 16 h.

Growth and assay of β-lactamase activity of E. coli transformants carrying in-frame β-lactamase fusion proteins

Escherichia coli JM105 lacking chromosomal β-lactamase was transformed with plasmids carrying in-frame β-lactamase fusions of FtsW domains, and grown at 37°C in LB containing 50 μg ml−1 kanamycin. β-Lactamase activity in the different cellular fractions was assayed using nitrocefin as substrate (O'Callaghan et al., 1972).

Polyclonal anti-FtsZ and anti-PBP3 antibodies

Polyclonal antibodies against purified recombinant FtsZ and PBP3 of M. tuberculosis were raised in rabbits by Imgenex, Bhubaneswar, India.

Tandem expression of FtsW and PBP3 and co-immunoprecipitation

Escherichia coli BL21 (DE3)/pJB307 cells (coexpressing His-FtsW and myc-PBP3) was induced with 25 μM IPTG at 25°C for 6 h. Cells were lysed with Cell Lytic™ B-II (Sigma Chemical, St Louis, MO, USA). The cell-free supernatant was incubated overnight with monoclonal anti-His antibody (1:100) (Roche Applied Science, Mannheim, Germany) at 4°C followed by addition of 5 μl of protein A/G-agarose. After incubation for 3 h at 4°C, the beads were pelleted, washed once in lysis buffer, boiled in SDS gel denaturing buffer, and separated on SDS-polyacrylamide gels prior to Western blotting with anti-myc antibody (Roche Applied Science).

Co-immunoprecipitation of FtsW, FtsZ and PBP3

Escherichia coli BL21(DE3) cells harbouring plasmids expressing S-tagged FtsZ (or its mutant) or myc-tagged PBP3 or His-tagged FtsW (G255–G524) separately, were induced with 100 μM IPTG for 3 h at 37°C. Cells obtained from 60 ml cultures were mixed, treated with 500 μl of Cell Lytic™ BII reagent for 15 min at room temperature. Three hundred microlitres of soluble cell lysate was precipitated with either S-protein agarose (10 μl) or with Ni2+-NTA agarose (10 μl) for 2 h at 30°C. The agarose-bound protein complexes were washed with 50 mM Na-phosphate, pH 7.5, 0.5 M NaCl and the protein samples were boiled with 20 μl of 2× Laemmli sample buffer for 10 min. Ten microlitres of sample was loaded on each lane and proteins were separated on SDS-PAGE followed by electrophoretic transfer to PVDF membrane. The blots were probed with monoclonal anti-His (1:1000) or anti-myc (Upstate, Charlottesville, VA, USA) (1:1000) or HRP-conjugated anti-S tag (EMD Biosciences, Novagen, Madison, WI, USA) (1:5000) antibody. Blots were subsequently incubated with HRP-conjugated rabbit anti-mouse IgG (1:1000) (Cell Signaling Technology, Beverly, MA, USA) in the case of probing with anti-His and anti-myc antibody and developed with Lumiglo (Cell Signaling Technology) chemiluminescence reagent.

Expression and purification of PBP3 and PBP3 (L314–T679) in E. coli BL21 (DE3)

Cells harbouring pJB301 or pJB303 were grown to an OD600 of 0.6. IPTG was added to a final concentration of 0.1 mM and growth was continued at 37°C with shaking for 4 h. Cells were lysed with Cell Lytic™ B-II (Sigma). His-PBP3 and -PBP3 (L314–T679) were purified from cell lysates by chromatography on Ni2+-NTA agarose.

Penicillin-binding assays

Purified His-PBP3 or His-PBP3 (L314–T679) was labelled with benzyl-[14C]-penicillin for 30 min at 37°C (Granier et al., 1994) and analysed by SDS-PAGE, Coomassie Blue staining and fluorography of the gels.

Pull-down assay

His-tagged proteins of different domains of FtsW were allowed to bind to Ni2+-NTA agarose. The soluble fractions of the cell lysate from E. coli cells expressing PBP3 (or its C-terminal domain) carrying a myc tag were incubated in each tube containing Ni2+-NTA agarose-bound FtsW-derived proteins for 1 h at 30°C. In each case, the slurry was washed thoroughly with PBS, boiled in SDS gel sample denaturing buffer, and loaded on SDS-polyacrylamide gels. The separated proteins were electroblotted onto nitrocellulose, blocked in blocking buffer [5% non-fat dry milk in 1× TBS containing 0.05% (v/v) Tween 20], and probed with anti-myc antibody. Detection was carried out by incubation with anti-mouse IgG-HRP-conjugate and enhanced chemiluminescence. In each case, blots were reprobed with anti-His antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) to ensure equal loading in all lanes.

Competitive inhibition of the association of protein PBP3 with His-FtsW (L361–A406) by peptide J4

This was studied by allowing myc-PBP3 to interact with His-FtsW (G255–G524) in 100 μl of buffer A [5 mM sodium phosphate buffer (pH 8), containing 120 mM KCl, 20 μg ml−1 phenylmethylsulphonylfluoride, 1 mg ml−1 gelatin] for 6 h at 25°C in the absence or presence of different concentrations of the peptide H2N-GLLPVTGLQLP-COOH (designated peptide J4) corresponding to the sequence of amino acid residues G376 to P386 of FtsW. A control reaction was set up with a scrambled peptide (H2N-LPQTGLPLGLV-COOH). Ni2+-NTA agarose was added to each tube, followed by incubation at 25°C. The suspension was centrifuged at 2000 g, the pellet was washed twice with 100 μl of buffer A, boiled for 5 min in SDS/PAGE-denaturing buffer and subjected to SDS/PAGE, followed by immunoblotting with anti-His antibody.

Analysis of the interaction between FtsW-derived peptide and PBP3 in solution

Biotinylated peptide J4 was incubated with PBP3 in buffer A for 6 h at 25°C in a volume of 100 μl. Control reactions were set up with an irrelevant (encoded by Rv0129c) His-tagged protein with pI similar to that of PBP3. After incubation, 10 μl (50% slurry) of streptavidin–agarose was added to each tube, followed by incubation for 30 min at 25°C. The suspension was centrifuged, the pellet was subjected to SDS-PAGE and immunoblotting as described above.

Plasmid constructions for the BACTH and BACTH complementation assays

Recombinant plasmids used in the BACTH complementation assays (Karimova et al., 2005) were generated by PCR amplification of the genes coding for the different Fts proteins using appropriate primers. A construct for expression of PBP3 was generated from pJB301, using the sense and antisense primers 5′-ATGGATCCCGGAAACGCGGTCATCTTGGTG-3′ (BamHI site in bold) and 5′-TATGAATTCTAGGTGGCCTGCAAGACCAAAGG-3′ (EcoRI site in bold) respectively, and subcloned into the corresponding sites of the pUT18C to generate pUT18C-PBP3. The construct for expression of FtsW was generated from pJB201, using the sense and antisense primers 5′-TCCATGGATCTAGAGTGCTAACCCGGTTGCTG-3′ (XbaI site in bold) (primer 1) and 5′-ATGGTACCACCCGTAACGCTGACCTT-3′ (KpnI site in bold) (primer 2) respectively, and cloning between the same sites of pUT18C and pKT25 to generate pUT18C-FtsW and pKT25-FtsW respectively. Mutants within the derivatives of FtsW were generated by overlap extension PCR using appropriate primers. The final round of PCR was performed using the sense and antisense primers 1 and 2 respectively. The products were cloned between the XbaI and KpnI sites (in bold) of pKT25 to generate mutants of FtsW in pKT25.

The resulting recombinant plasmids expressed hybrid proteins in which the polypeptides of interest were fused to the C termini of two fragments T25 (for FtsW or its mutants) and T18 (for FtsW and PBP3) of the catalytic domain of Bordetella pertussis adenylate cyclase (AC).

Full-length ftsZ gene was PCR-amplified from pJB101 using the primer pair 5′-TTTTAAGCTTTATGACCCCCCGCACAACTA-3′ (sense) and 5′-ATGGATCCAGATATCGCGGCGCATGAAGGGCGGCA-3′ (antisense) (primer 3) and cloned between the HindIII and BamHI (sites in bold) of pKNT25 generating pKNT25-FtsZ, where FtsZ is fused to the N-terminal end of the T25 fragment of AC. The mutant of FtsZ deleted in residues D367 to D370 was generated by overlap extension PCR essentially as described before (Datta et al., 2002) and the final product was cloned between the HindIII and BamHI sites of pKNT25.

For coexpression of FtsW and FtsZ, the ftsW and ftsZ genes were cloned in the plasmid pCKB110 (Choudhuri et al., 2002). Briefly, pJB101 was digested with NdeI and HindIII, the fragment encoding ftsZ (or its mutant) gene was excised and recloned into the same sites downstream of the Shine–Dalgarno sequence present in pCKB110 to generate pJB111. The ftsW gene was amplified from pJB201 using the primers 1 and 2, and cloned between the NcoI and KpnI sites (in bold) upstream of the Shine–Dalgarno sequence in pJB111, to generate pJB112 harbouring the ftsW and ftsZ genes with the artificial ribosome binding site AGGA upstream of the ftsZ gene. Subsequently the cassette containing the ftsW and ftsZ genes was amplified from pJB111 by PCR using the sense and antisense primers 1 and 3 respectively, and cloned between the XbaI (in bold) and SmaI sites of pKT25 to generate the pKT25-ftsW/ftsZ plasmid containing the ftsW and ftsZ genes fused to the T18 gene of AC. The mutated versions of the ftsW or ftsZ genes were generated by overlap extension PCR as described earlier.

For BACTH complementation assays, recombinant pKT25 (or pKNT25) and pUT18C carrying fts genes were used in various combinations to co-transform E. coli DHM1 cells (Karimova et al., 2005). β-Galactosidase activities of the transformants were determined as described by Karimova et al. (2005).

Construction of FtsW replacement vectors in M. smegmatis

A two-step homologous recombination strategy as described by Parish and Stoker (2000) was used to disrupt M. smegmatis ftsW at its native locus in the presence of an integrated copy of either wild-type or mutant ftsW of M. tuberculosis. Two vectors p2NIL, the gene manipulating vector without any mycobacterial origin of replication and pGOAL17, the marker gene cassette-containing vector were used for this purpose. A suicide plasmid with an 840 bp internal deletion in the ftsW gene was constructed in two steps. In the first step, a 779 bp DNA fragment encompassing the 5′ end of ftsW and its upstream flanking region was PCR amplified using the primer pair 5′-AAAAAAGCTTACGCGTTATCGCGACACG-3′ (sense) and 5′-TAGGTACCAGCCGCGTGAACCGTTGG-3′ (antisense). The amplicon was cloned in p2NIL between the asymmetric HindIII and KpnI sites (bold) to give rise to pJB398. In the next step, 736 bp of DNA fragment bearing the 3′ end of ftsW and its downstream flanking region was amplified using the primer pair 5′-ATGGATCCTCGCAGGCGACCAC-3′ (sense) and 5′-TGGTACCAGTTCCAGGTCATAGCCG-3′ (antisense). The amplicon was cloned in pJB398 between the BamHI and KpnI sites (bold) to give rise to pJB399. Finally, a 6.1 kb PacI fragment carrying the lacZ, aph and sacB genes was isolated from pGOAL17 and inserted into pJB399 to generate the suicide plasmid pJB401.

Isolation of the strains carrying wild-type or mutated ftsWMTB

Mycobacterium smegmatis mc2 155 was electroporated with denatured pJB401 DNA. SCOs were selected on agar plates containing kanamycin and Xgal. Kanamycin-resistant, blue, SCO were isolated. In the next step, His-tagged ftsWMTB (or its mutants) were cloned under the control of the hsp60 promoter as described earlier (Dasgupta et al., 2006) and hsp60-ftsW was cloned in pUC19. A 3.757 kb Hyg-integrase cassette was excised from pUC-HY-INT (Mahenthiralingam et al., 1998) and inserted into the recombinant pUC construct to generate an integrative vector. In order to inactivate ftsW at its native location, ftsWMTB or its mutants was first cloned in the integrative vector as described above, and electroporated into the SCO. The resultant merodiploid strains were screened for DCOs as described by Chauhan et al. (2006). White, Kans, Hygr and sucrose-resistant DCO colonies were analysed by PCR as well as by Southern blotting using primers selective for ftsW of M. tuberculosis or M. smegmatis. The expression of FtsWMTB or its mutants in DCO was also confirmed by Western blotting using anti-His-antibody. The viability as well as growth kinetics of DCOs were analysed as described earlier (Dasgupta et al., 2006).

Analysis of interactions among FtsZ, FtsW and PBP3 in mycobacteria

Mycobacterium smegmatis mc2 155 or its transformants were lysed using 0.1 mm zirconia glass beads and a Mini Bead beater (Biospec Products, Oklahoma) (Yamamoto et al., 2001) and centrifuged at 10 000 g for 15 min. Membranes were pelleted from the supernatant by centrifugation at 100 000 g for 1 h. Membranes were solubilized with 2% Triton X-100 in 10 mM Tris-HCl buffer, pH 8.0 at 4°C. Solubilized membranes based on equivalent cell number or protein concentration were immunoprecipitated with either anti-His, or anti-PBP3, or anti-FtsZ antibody or with pre-immune sera followed by Western blotting with appropriate antibodies.

Conditional inactivation of ftsW in mycobacteria

For conditional inactivation of ftsW, antisensing of ftsW was performed by PCR amplification of the ftsW gene using the primer pair 5′-ATGGATCCTCACCCGTAACGCTGACCTTC-3′ and 5′-ATGAATTCGTGGGCAGCATCCTGACC-3′, followed by cloning of the PCR product in reverse orientation in the vector pMIND (Blokpoel et al., 2005) between the BamHI (in bold) and EcoRV sites to generate pJB223.

In order to conditionally inactivate ftsW, M. smegmatis mc2 155 was electroporated with either pJB223 or the empty vector. Transformants were grown up to mid log phase and induced with tetracycline (20 ng ml−1) for different periods of time up to 24 h. Northern analysis was performed as described by Ji et al. (1999) to confirm antisensing of ftsW. Briefly, M. smegmatis was grown in the absence or in the presence of tetracycline, and total RNA was extracted by using a Qiagen RNeasy mini protocol kit (Qiagen GmbH, Hilden, Germany). Ten microgram aliquots of RNA were separated by electrophoresis on a 1.2% agarose−1.8% formaldehyde gel and blotted onto a nylon membrane (Amersham Biosciences, a division of GE Healthcare, Buckinghamshire, UK). RNA was cross-linked to the membrane by UV irradiation by using a UV Stratalinker (Stratagene, Germany). Single-stranded DNA oligonucleotides probes for either sense ftsW RNA, 5′-CGAGGATCCACCGGCAGAGATGAGCGGCAGCTACAG-3′) or antisense ftsW RNA, 5′-TGATCATGGTGCTCTCGGCGTCGGGCG-3′ or 16s RNA, 5′-GCGATTACTAGCGACGCCGACTT-3′ were labelled using T4 polynucleotide kinase according to standard procedures. Blots were pre-hybridized and then hybridized with [32P]-labelled single-stranded DNA oligonucleotides (100 pmol). The DNA–RNA hybridization was detected after exposure to X-ray film.

Fluorescence microscopy

Immunostaining was adapted from the method of Harry et al. (1995) with some modifications. Cells were fixed by incubation for 15 min at room temperature followed by 45 min on ice in 2.5% (v/v) paraformaldehyde, 0.04% (v/v) glutaraldehyde, 30 mM sodium phosphate (pH 7.5). After washing in PBS, the cells were permeabilized by exposing to 2% toluene for 2 min, and immediately transferred to slides. The slides were washed with PBS, air-dried, dipped in methanol (−20°C) for 5 min and then in acetone (−20°C) for 30 s and allowed to dry. After rehydration with PBS, the slides were blocked for 2 h at room temperature with 2% (w/v) BSA-PBS and incubated for 1 h with appropriate dilutions of primary antibody in BSA-PBS. The slides were washed extensively with PBS and then incubated with a 1:1000 dilution of Alexa 488-conjugated anti-rabbit IgG (Molecular Probes, Eugene, OR, USA) in BSA-PBS. For staining of DNA, 2 μg ml−1 of 4′,6-diamino-2-phenylindole (DAPI) was included with the secondary antibody. WGA-Alexa 488 (Molecular Probes) 2 μg ml−1 was used for staining of cell walls (Sizemore et al., 1990). WGA is a lectin that binds to oligomers of N-acetylglucosamine and N-acetylmuramic acid. After extensive washing with PBS, the slides were mounted using 50% glycerol. In controls for assessing specificity of the primary antibodies, the incubation with the pre-immune sera was included.

Acknowledgements

This work was supported in part by grants from the Indian Council of Medical Research and the Department of Biotechnology to J.B. The authors would like to thank Dr Stewart Cole, Institut Pasteur, Paris for the cosmid MTCY270, Dr Daniel Ladant and Dr Gouzel Karimova, Institut Pasteur, Paris for the reagents for the BACTH assays, Dr Brian Robertson, Imperial College, London for the vector pMIND, Dr Richard Stokes, University of British Columbia, Vancouver for the vector pUC-HY-INT, Neil Stoker, Royal Veterinary College, London, and Tanya Parish, Queen Mary's School of Medicine and Dentistry, London, for the vectors p2NIL and pGOAL17, Dr Anuradha Lohia, Bose Institute for fluorescence microscopy and Dr Martine Nguyen-Disteche, University of Liege, Belgium, for helpful discussions.

Ancillary