Summary Three membrane proteins required for cell division in Escherichia coli, FtsQ, FtsL and FtsB, localize to the cell septum. FtsL and FtsB, which each contain a leucine zipper-like sequence, are dependent on each other for this localization, and each of them is dependent on FtsQ. However, FtsQ is found at the cell division site in the absence of FtsL and FtsB. FtsQ, in turn, requires FtsK for its localization. Here, we show that FtsL, FtsB and FtsQ form a complex in vivo. Strikingly, this complex forms in the absence of FtsK, which is required for the localization of all three proteins to the mid-cell. These findings indicate that the FtsL, FtsB, FtsQ interactions can take place in cells before movement to the mid-cell and that migration to this position might occur only after the formation of the complex. Evidence indicating the regions of the three proteins involved in complex formation is presented. These findings provide the first example of preassembly of a subcomplex of cell division proteins before their localization to the septal region.
Bacterial cell division involves a large number of proteins that co-ordinate invagination of the cell membrane, inward growth of the peptidoglycan layer, constriction of the outer membrane and, finally, separation of daughter cells. In Escherichia coli, cell division requires the assembly of at least 11 proteins at the mid-cell (Buddelmeijer and Beckwith, 2002; Bernhardt and de Boer, 2003). The precise function of only a few of the cell division proteins is known. FtsZ, a tubulin homologue with GTPase activity, forms a ring at the mid-cell and is thought to initiate the formation of the septal complex of proteins and to be important for the contraction of the septal ring during the invagination process (Romberg and Levin, 2003). The FtsI protein (a penicillin-binding protein) exhibits an enzymatic activity involved in peptidoglycan synthesis required specifically for cell division (Botta and Park, 1981). The role of the remaining proteins is either poorly defined or not understood at all.
A ‘pathway’ for assembly of these 11 proteins at the mid-cell has been proposed, based on studies of the localization of each protein under conditions in which cells are depleted for each of the other proteins. Formally, this assembly process appears to be a sequential one in which, for the most part, one protein is dependent for its localization on the presence of the preceding protein at the mid-cell (Fig. 1).
There are two exceptions to the sequential dependency pathway. First, the proteins ZipA and FtsA, each of which contributes to the stability of the previously formed FtsZ rings (Pichoff and Lutkenhaus, 2002), are dependent on FtsZ for their mid-cell positioning, but are not dependent on each other (Hale and de Boer, 1999; Liu et al., 1999). Secondly, the proteins FtsL and FtsB (previously named YgbQ), although dependent on FtsQ and earlier proteins in the pathway, are also dependent on each other for localization (Buddelmeijer et al., 2002).
The localization and function of cell division proteins has also been studied extensively in the Gram-positive bacterium Bacillus subtilis. Unlike in E. coli, the B. subtilis cell division proteins do not localize to the septal region in a strict sequential pathway, and the membrane-bound proteins in particular seem to localize in a co-operative fashion (Daniel et al., 1998; Daniel and Errington, 2000; Katis et al., 2000).
FtsL and FtsB of E. coli are small bitopic transmembrane proteins, each with a single cytoplasmic, transmembrane and periplasmic domain. They both contain leucine zipper-like sequences that comprise a major portion of their small (64 residues and 81 residues respectively) periplasmic domains. The stability of FtsL appears to be dependent on the presence of FtsB (Buddelmeijer et al., 2002). The co-dependency of FtsL and FtsB for their localization, the fact that they both contain potential leucine zippers and the destabilization of FtsL in the absence of FtsB suggested that these two proteins might form heterodimers or higher order oligomers.
In this paper, we show that FtsL and FtsB form a complex in vivo. In strains depleted for FtsQ, the protein that precedes these two in the pathway, FtsL and FtsB are no longer found in a complex. Because we find that FtsQ is also co-precipitated with the FtsL–FtsB complex, our results suggest that FtsQ is required for the stable interaction of FtsL and FtsB. Finally, we find that the FtsQ–L–B complex is formed in a strain depleted for FtsK, a protein further upstream in the assembly pathway. Together, our findings suggest that the formation of the FtsQ–L–B complex is not dependent on the localization of any of the components of the complex to the mid-cell. Recent findings in E. coli, B. subtilis and the yeast Saccharomyces cerevisiae allow us to compare the process of localization and interaction of division proteins in these very different organisms.
FtsL and FtsB exist as a complex in vivo
To test whether FtsL and FtsB indeed interact in vivo, we performed co-immunoprecipitation experiments on protein fractions from Triton X-100-solubilized membranes (Experimental procedures). As antibodies to FtsL were available, but antibodies against FtsB were not, FtsB was tagged at the carboxy-terminus with a FLAG epitope. This version of FtsB can complement an ftsB depletion strain (data not shown). To maintain this latter construct at low copy number, the ftsB-flag3 fusion was integrated at the chromosomal lambda attachment site using λInCh (Boyd et al., 2000). The resulting strain (NB1050) is merodiploid for ftsB as, in addition to the ftsB-flag3 copy, a wild-type copy of ftsB is present at its original locus at 62 min on the chromosome. Expression of ftsB-flag3 from λInCh is under the control of the pBAD (arabinose-inducible) promoter. The FtsL detected in these experiments is expressed from a wild-type copy of the ftsL gene located at its normal position on the chromosome and expressed from its natural promoter.
Immunoprecipitates of extracts made from strain NB1050 grown in the presence of arabinose using either α-FtsL or α-FLAG antibody were subjected to Western blot analysis (Fig. 2, lanes 7–9). We found that the precipitates with α-FtsL antibodies contain both FtsL and FtsB-Flag. The Western analysis also showed that α-FLAG antibodies precipitated both FtsB-Flag and FtsL from arabinose-induced extracts. In the absence of arabinose and the presence of glucose, FtsB-Flag is no longer expressed, and only FtsL is detected (Fig. 2, lanes 10–11). Preimmune serum failed to precipitate either protein. These results suggest that FtsL and FtsB form a complex in vivo.
Quantification of the amount of FtsL and FtsB-Flag3 that was precipitated with α-FtsL and α-FLAG antibodies indicates that the precipitation of the proteins is efficient. Approximately 75% of FtsL and 90% of FtsB-Flag3 was found in the precipitated complex when α-FtsL and α-FLAG antibodies were used respectively. Precipitation of FtsB was more efficient than precipitation of FtsL, most probably because the FLAG epitope is more accessible to binding with the anti-FLAG antibodies as it is attached to the C-terminal domain of the protein. Antigenic regions in FtsL that are recognized by the α-FtsL antibodies might be masked by direct protein–protein interactions.
In results from various experiments reported here, we occasionally saw small amounts of FtsB-Flag precipitated by α-FtsL when cells were depleted of FtsL (see Fig. 6). These repeated findings might indicate that there is a slight cross-reactivity of FtsL antibody with FtsB, perhaps because both proteins contain leucine zipper-like motifs. Alternatively, the preparation of FtsL to elicit antibody, which was a version of FtsL with a His6-tag, might have contained some FtsB after purification on the nickel column. However, this seems unlikely as FtsB-Flag was not detected with antibodies against FtsL on Western blot (Fig. 3). Finally, in some cases, a small amount of FtsL, too low to be detected with α-FtsL antibodies, might remain under the depletion conditions, and this amount might have been sufficient to allow some FtsB-Flag to be precipitated and detected. Furthermore, a triple FLAG tag was fused to FtsB, which allows detection of very small amounts of this protein by α-Flag antibodies.
We have reported previously that the absence of FtsB reduces the amount of FtsL to the point at which it can no longer be detected on Western blots, perhaps as a result of destabilization (Buddelmeijer et al., 2002). This finding explains the failure to detect FtsL in immunoprecipitation experiments done with a strain depleted for FtsB (Fig. 2, lanes 12–14). We asked whether the converse would be the case: are FtsB levels reduced in the absence of FtsL? We found that, when FtsL was depleted from cells, significant amounts of FtsB-Flag were still detected on Western blots (data not shown), suggesting that the amounts of FtsB are not affected in the absence of FtsL. This conclusion should be considered tentative, however, as the FLAG tag might alter the stability of FtsB-Flag.
We were concerned that the co-immunoprecipitation of FtsL and FtsB might not reflect the actual existence of a complex between the two proteins. For instance, proteins of the cytoplasmic membrane might tend to co-immunoprecipitate even if they are not interacting in the cells. To test for such an artifact, we asked whether another completely unrelated membrane protein might also be found in such precipitates. We chose a cytoplasmic membrane protein (CcmG), the topology and size of which are similar to those of FtsL and FtsB, as a control to test for the specificity of their interactions. CcmG is involved in an electron transfer process required for cytochrome c biogenesis (Reid et al., 2001) and is not required for cell division. CcmG carrying a histidine tag at its carboxy-terminus was overproduced from a pBAD promoter on a plasmid in strains carrying a second plasmid expressing either FtsL or FtsB-Flag or with no additional plasmid. Note that, as the antibodies against FtsL were raised against FtsL-His6 (Ghigo et al., 1999), both FtsL and CcmG-His6 are precipitated with α-FtsL antibodies (Fig. 4). However, only CcmG-His6 and not FtsL is precipitated with α-His6 antibodies (Fig. 4). Thus, under the same conditions in which FtsL and FtsB co-precipitate, FtsL and CcmG do not. Similar results were obtained when FtsB-Flag and CcmG-His6 were overproduced in the same strain. The precipitation of CcmG-His6 with α-His6 did not carry with it any FtsB-Flag, and the FtsB-Flag precipitated with α-FLAG showed no co-precipitated CcmG-His6 (Fig. 4). These results strengthen our conclusion that the FtsL–FtsB interaction suggested by our experiments reflects the existence of an in vivo FtsL–FtsB complex.
The transmembrane domain and the periplasmic leucine zipper-like domain, but not the cytoplasmic domain of FtsL, are required for the interaction with FtsB
We wished to determine which regions of FtsL and FtsB are involved in the interactions between the two proteins. FtsL and FtsB have a short cytoplasmic domain, a transmembrane segment and a periplasmic domain that exhibits a leucine zipper-like sequence. Previous studies from this laboratory on the functioning of FtsL have used constructs of FtsL in which one or another of these domains was exchanged for a similar domain from another protein. We call such constructs ‘swap constructs’ or ‘swaps’. These studies suggested that all three domains of FtsL are required for its function, but that the cytoplasmic domain is not required for localization to the septum (Ghigo and Beckwith, 2000). Here, we use these swap constructs to determine the features of FtsL that are important for its interaction with FtsB.
The swap constructs are derived from a version of wild-type FtsL that contains restriction sites bordering the sequence coding for the transmembrane segment in the ftsL gene. This version of FtsL is termed LLL, as we name these constructs by the source of the domains that they carry, starting at the amino-terminus of the protein. When we substitute the amino-terminal cytoplasmic domain of the MalF protein for that of FtsL, the construct is designated FLL, etc. (MalF is a maltose transport protein that does not have a role in cell division).
To identify the domains in FtsL that are important for the interaction with FtsB, immunoprecipitations were performed with strains that were depleted for the wild-type copy of FtsL but were expressing a swap construct of ftsL from a Lac promoter on a low-copy plasmid. The ftsB-flag3 is expressed from the chromosome at the lambda attachment site from a modified pTrc promoter that can be induced by IPTG. The LLL construct, which effectively complements an FtsL depletion strain, behaves like the wild-type FtsL in co-immunoprecipitation experiments, i.e. antibodies against FtsL or FLAG bring down LLL and FtsB-Flag together (data not shown). When the FLL swap was expressed from the low-copy plasmid, the hybrid protein was also precipitated by both α-FtsL and α-FLAG antibody (Fig. 5). This result indicates that the cytoplasmic domain of FtsL is not required for the interaction with FtsB.
With the constructs LFL and LQL, the membrane-spanning domain of FtsL was replaced with the transmembrane domains of MalF and FtsQ respectively (Guzman et al., 1997). These FtsL swap proteins do not complement an FtsL depletion strain and are thus not functional in cell division (Guzman et al., 1997; Ghigo and Beckwith, 2000). LFL is produced at levels slightly higher (about 1.5-fold) than wild-type levels of FtsL, in both the membrane and the solubilized membrane fraction (data not shown). Levels of LQL are approximately twofold higher and LLL is present at about threefold higher levels than those of FtsL (data not shown). Our antibody to FtsL precipitates both LFL and LQL (Fig. 5). In the case of LFL, FtsB-Flag was not co-precipitated. Reciprocally, α-FLAG antibody precipitated FtsB-Flag but not LFL; however, when α-FtsL was used to precipitate LQL, this antibody did bring down some FtsB-Flag. In the reciprocal experiment, α-FLAG antibody co-precipitated a small amount of LQL along with FtsB-Flag (Fig. 5). The results with the LFL construct indicate that the transmembrane domain of FtsL is important for the interaction with FtsB. However, although this domain could not be replaced with the transmembrane domain of MalF, the membrane-spanning sequence of FtsQ still allows some interaction of FtsB. It is possible that, as FtsQ itself is part of the division machinery, it contributes somewhat to the stabilization of the interaction between FtsL and FtsB through its membrane-spanning segment (see below for results on interactions between FtsQ, L and B). Alternatively, the nature of the MalF transmembrane segment might affect the overall structure of the FtsL protein in some other way.
Studies with swap constructs showed that the leucine zipper-like motif of FtsL is involved in localization of this protein to the division site (Ghigo and Beckwith, 2000). In the FtsL swap protein LLhinfL, the leucine zipper-like motif of E. coli FtsL is replaced with the corresponding region of the Haemophilus influenzae FtsL, but the carboxy-terminal portion of the periplasmic domain derives from E. coli FtsL. This protein localizes to the constriction site but less efficiently than E. coli FtsL and complements only very weakly the cell division defect of an E. coli strain depleted for FtsL (Ghigo and Beckwith, 2000). Using this construct in the immunoprecipitation experiments, we can determine whether the leucine zipper-like motif plays a role in the interaction with FtsB. LLhinfL is recognized by the antibody against E. coli FtsL. Levels of LLhinfL in the solubilized membrane fractions were similar to those found with LLL in earlier experiments (data not shown). Although LLhinfL was immunoprecipitated by α-FtsL antibodies, only a faint band of FtsB-Flag was detected in the precipitate (Fig. 5). In the reciprocal experiments, low amounts of LLhinfL were detected when α-FLAG was used to precipitate FtsB-Flag (Fig. 5). Thus, compared with LLL or FLL, LLhinfL interacts weakly with FtsB, indicating that the leucine zipper-like motif of FtsL is a major contributor to the interaction. As the arrangement of leucine residues is conserved in the leucine zipper-like motifs of E. coli and H. influenzae, other residues in this domain are involved in the interaction with FtsB.
The results presented in this section suggest that the transmembrane segment and the leucine zipper-like motif of FtsL are important for its interaction with FtsB, but that the cytoplasmic domain is not. The specificity of the effects seen with the several different swap constructs further supports our conclusion that the FtsL–FtsB interactions reflect the presence of a complex in vivo.
FtsL and FtsB can interact before their localization to the mid-cell position
The 11 cell division proteins of E. coli localize to the division site mostly in a sequential order, although FtsL and FtsB are dependent on each other for proper localization. In this sequential order, both FtsL and FtsB depend on FtsQ for their localization, and all three proteins, in turn, are required for the localization of FtsW and FtsI (Chen and Beckwith, 2001; Buddelmeijer et al., 2002). In order to study whether complex formation between FtsL and FtsB can occur before or in conjunction with localization to the division site, we analysed the interaction between FtsL and FtsB in the absence of either of the other two cell division proteins, FtsQ and FtsI, that localize upstream or downstream of FtsL and FtsB respectively. The ftsB-flag3 construct was introduced at the lambda attachment site under the control of a modified pTrc promoter (IPTG-inducible) in strains that could be depleted for either FtsQ or FtsI. In the presence of arabinose, either FtsQ or FtsI was expressed from a pBAD promoter on a complementing plasmid. As FtsL is unstable in the absence of FtsB, we were concerned that other Fts proteins might exhibit instability in strains depleted for other members of the pathway. Therefore, we first determined the protein levels of the different Fts gene products when cells were depleted for one of the other gene products. We found that the relative steady-state levels of FtsL were not affected in the absence of FtsQ and FtsI, that the steady-state levels of FtsQ were not affected in the absence of FtsL and FtsI, and that the steady-state levels of FtsI were not affected in the absence of FtsQ and FtsL (Supplementary material, Fig.S1).
We then proceeded to deplete cells of either FtsI or FtsQ and assess the FtsL–FtsB interaction by immunoprecipitation. When FtsQ or FtsI was expressed from the arabinose promoter in strains carrying chromosomal null mutations for either of the two genes, FtsL and FtsB-Flag were precipitated with α-FtsL and α-FLAG (Fig. 6). These results are consistent with those in earlier figures. When cells were depleted for FtsQ or FtsI by removing arabinose from the media and adding glucose, cell division was inhibited, and extensive filamentation was observed, as expected (data not shown). Under these conditions, we still found co-immunoprecipitation of FtsL and FtsB-Flag in the FtsI depletion strain, but not in the FtsQ depletion strain (Fig. 6). As FtsQ depletion did not deplete wild-type FtsL from the membrane (Supplementary material, Fig.S1), the weak FtsB-Flag band is probably caused by a weakened interaction between FtsL and FtsB in the absence of FtsQ. The presence of small amounts of an FtsB-Flag band in the FtsL-depleted cells when FtsL was immunoprecipitated (Fig. 6) might reflect the phenomenon that we have described earlier in this paper.
These results indicate that FtsQ, but not FtsI, is required for the interaction between FtsL and FtsB and are consistent with proposals in which either the interaction between FtsL and FtsB occurs after the proteins follow FtsQ to the mid-cell position, or the three proteins form a complex before their localization to the septal region. To analyse these possibilities, we used an FtsK depletion strain to see whether FtsL and FtsB are able to interact without FtsK. We have shown previously that neither FtsL nor FtsQ localizes to the mid-cell in the absence of FtsK (Chen and Beckwith, 2001). If the interaction between FtsL and FtsB is dependent on their localization to this site, then that interaction should be abolished in the absence of FtsK. Growth of an FtsK (pBAD-controlled) depletion strain in the presence of glucose results in extensive filamentation, conditions under which we know that FtsQ and FtsL (and almost certainly FtsB) do not localize to the septum (Chen and Beckwith, 2001). Furthermore, FtsK was detected on a Western blot in total-cell lysates of an FtsK depletion strain grown in the presence of arabinose when a complementing copy of ftsK was expressed, but not in extracts obtained from filaments grown in the presence of glucose (Fig. 7). Under depletion growth conditions, the reciprocal experiments with α-FtsL and α-FLAG antibodies showed that FtsL and FtsB-Flag were co-immunoprecipitated (Fig. 6). These results suggest that FtsK is not necessary for the interaction between FtsL and FtsB. As neither FtsL nor FtsQ is localized to the mid-cell in the FtsK depletion conditions (Chen and Beckwith, 2001), our findings suggest that the FtsL–FtsB complex can be formed before its localization to the division site and that the proteins might localize to the mid-cell as a preassembled complex.
FtsQ is part of the FtsL–FtsB complex
As the interaction between FtsL and FtsB is independent of prior localization to the division site, but the formation of that complex is dependent on FtsQ, FtsQ might be part of the FtsL–FtsB complex. In order to test whether FtsQ interacts with the FtsL–FtsB complex, immunoprecipitations were performed using polyclonal antibodies against FtsQ.
We first tested whether FtsQ could be precipitated from Triton X-100-solubilized membranes using polyclonal antibodies against FtsQ. When FtsQ was overproduced from a pBAD promoter and subjected to precipitation with antibodies against FtsQ, FtsQ was detected on Western blots probed with monoclonal antibodies against FtsQ (Buddelmeijer et al., 1998). FtsQ was not detected in immunoprecipitates using preimmune serum or in immunoprecipitates using FtsQ antibodies with extracts from membranes that are depleted of FtsQ, indicating that FtsQ was precipitated specifically with α-FtsQ antibodies (Fig. 8A). Furthermore, the polyclonal antibodies against FtsQ did not cross-react with either FtsL or FtsB-Flag (Fig. 3).
To test whether FtsQ is part of the FtsL–FtsB complex, immunoprecipitations were performed on membrane extracts from a strain that expresses endogenous FtsQ and FtsL from the chromosome and FtsB-flag3 from a modified pTrc promoter that is integrated into the chromosome at the lambda attachment site. FtsQ was detected on a Western blot with monoclonal antibodies against FtsQ. FtsQ was precipitated with both α-FtsQ antibodies and α-FLAG antibodies (Fig. 8B), indicating that FtsQ interacts directly or indirectly with FtsB. FtsL was co-precipitated with α-FtsQ antibodies; however, FtsQ was not co-precipitated with α-FtsL antibodies. The failure to detect FtsQ in the precipitate with α-FtsL might have resulted from the low levels of FtsQ in these strains, which are close to the detection limit of the Western blot analysis. Comparison of FtsL levels in precipitates with α-FtsL or α-FLAG and α-FtsQ also suggests that less FtsL protein is co-precipitated with α-FtsQ (Fig. 8B). FtsL and FtsB-Flag were co-precipitated, as shown previously.
To facilitate detection of FtsQ after precipitation, the protein was tagged at its carboxy-terminus with a triple c-Myc tag and expressed from a high-copy pBAD plasmid. Expression of FtsB-flag3 from the chromosome is under the control of a modified pTrc promoter (IPTG-inducible) while FtsL is expressed from its natural promoter. We found that the immunoprecipitates with α-FtsQ antibodies contain FtsQ-Myc, FtsL and FtsB-Flag (Fig. 8C). The Western analysis also showed that α-FLAG precipitated FtsB-Flag, FtsQ-Myc and FtsL and that α-FtsL precipitated FtsL, FtsB-flag and FtsQ-Myc. These results indicate that FtsQ, FtsL and FtsB form a complex in vivo.
Comparison of the intensities of the FtsL bands in the immunoprecipitates with either α-FtsQ or α-FLAG antibodies showed that more FtsL was co-precipitated with α-FtsB than with α-FtsQ. Similarly, FtsQ was more efficiently co-precipitated with α-FtsB than with α-FtsL. We are not sure of the significance of these findings. They might simply represent differing effects of the antibodies on the interactions involved in complex stability or expression levels of proteins. We cannot conclude which direct interactions exist between these proteins, but the results do show that the three proteins form a complex in vivo.
Studies on the localization to the mid-cell of proteins involved in cell division in E. coli have raised the possibility that there is a sequential pathway of assembly of these proteins at the septal region (Buddelmeijer and Beckwith, 2002). The results presented here suggest that this formalistic representation of a series of sequential interactions at the septal region might not reflect the actual situation.
Using co-immunoprecipitation experiments, we show that FtsL and FtsB form a complex in vivo. We also find that FtsQ is co-precipitated with FtsL and FtsB and that it is, in fact, required for the interaction between the latter two proteins. Strikingly, this multimeric protein complex forms in the absence of FtsK, a protein required for the localization of all three proteins to the mid-cell (see Fig. 1). These results indicate that FtsQ, FtsL and FtsB are capable of oligomeric interactions before movement to the mid-cell and that migration to this position might occur only after formation of the complex. As the three proteins are present in the cell before they are seen at the septal region, it is the complex of proteins, in all likelihood, that is recruited to the mid-cell during the division cycle.
We have used two other methods to study the interactions of cell division proteins (data not included). We performed affinity chromatography using a Ni2+ column to which His6-FtsL could be specifically bound and found that His6-FtsL eluted from the column together with FtsB-Flag, whereas aspecifically bound proteins were identified in the void before elution with imidazole (data not shown). We are using a bacterial two-hybrid system based on the lambda repressor headpiece dimerization assay (Leeds and Beckwith, 1998) to study the interaction between various membrane-bound cell division proteins. These studies showed that FtsL, FtsB and FtsQ interact in vivo (M. Gonzalez, unpublished results). Furthermore, data obtained with different bacterial two-hybrid systems also showed an interaction between FtsL, FtsB and FtsQ (DiLallo et al., 2003; D. Ladant, personal communication).
We have not yet studied whether other proteins may be part of the FtsQ, FtsL, FtsB complex. The finding that FtsK localizes in the absence of these proteins and is not required for their assembly into an oligomeric complex might indicate that this protein is not a component of this assemblage. On the other hand, FtsQ, despite being co-precipitated with FtsL and FtsB, does not require these proteins for its appearance at mid-cell. Future analogous studies with the remaining proteins in the pathway should indicate whether a more extensive complex exists.
FtsQ can localize in the absence of its complex with FtsB and FtsL, but the reverse is not true. It appears that FtsQ is required for the interaction between FtsB and FtsL and that interaction is necessary for localization of the latter two proteins.
One result is somewhat surprising. Here and previously, we have shown that FtsL is reduced in amounts and likely to be unstable in the absence of FtsB. We might therefore expect that, in an FtsQ depletion strain, where the FtsL–FtsB complex is not observed in immunoprecipitation experiments, FtsL would be present in reduced amounts. This is not the case. Our results suggest that FtsL and FtsB might still have some affinity, albeit reduced, in the absence of FtsQ, and that this weakened interaction is still sufficient to protect FtsL from degradation. Alternatively, FtsL might be protected from proteolysis when not localized to the mid-cell.
FtsL and FtsB contain leucine zipper-like motifs in their periplasmic domains and, at least for FtsL, this motif is a major contributor to the interaction between FtsL and FtsB. The transmembrane domain of FtsL is also important for the formation of the complex. Secondary structure predictions indicate that the α-helix of the leucine zipper-like motif extends into the membrane; thus, the membrane-spanning segment and the leucine zipper-like motif might together permit the formation of a coiled-coil structure in oligomeric complexes. The fact that FtsL is destabilized in the absence of FtsB might indicate that the formation of a coiled-coil structure between the leucine zipper-like motifs is essential for its stabilization.
The only region of FtsQ that is essential for its localization to the septum and for its function there is its periplasmic domain (Chen et al., 1999). As long as FtsQ is anchored to the membrane, independent of the source of the membrane anchor, cell division proceeds normally or relatively normally (Guzman et al., 1997). However, FtsQ does not contain a leucine zipper-like motif in its periplasmic domain, indicating that its interaction with FtsL and FtsB probably does not involve a leucine zipper. We have described previously mutations of ftsQ that alter the protein so that it still localizes to the mid-cell, but is no longer able to recruit FtsL to the region (Chen et al., 2002). These mutant FtsQ proteins, altered near the carboxy-terminus of the protein, might define a part of the protein that is involved in the interaction with FtsL and FtsB. This same region shows up as surface exposed in structure predictions for FtsQ, consistent with its possible role in protein–protein interactions (Buddelmeijer et al., 1998).
Studies on the comparable process of cell division in other bacteria and budding yeast reveal similarities in the assembly of cell division proteins at the division plane. The Gram-positive bacterium B. subtilis expresses several proteins that are homologues of E. coli cell division proteins. These include FtsL, DivIC (a probable relative of FtsB; Buddelmeijer et al., 2002) and DivIB (a homologue of FtsQ). In the absence of DivIB, FtsL is destabilized, suggesting a direct interaction between these two components (Daniel and Errington, 2000). However, DivIC (FtsB homologue) is not affected under these conditions (Katis et al., 2000). Although depletion of FtsL results in a rapid decrease in the amount of DivIC (Daniel et al., 1998) (DivIB is not affected), depletion of DivIC does not result in loss of FtsL (Daniel and Errington, 2000). The finding that the cell division proteins FtsL, DivIC and DivIB of B. subtilis localize to the septum in an interdependent fashion and, in some cases, require the presence of each other for stability suggests that these homologues of the three E. coli proteins studied here may also exist as a complex. Although the individual proteins of E. coli and B. subtilis are not highly conserved (15% identical residues in both FtsL and FtsB/DivIC) and the pattern of stability of these proteins is different, the general relationship seems to be similar. Recently, Robson et al. (2002) postulated that FtsL and DivIC interact indirectly through binding to other cell division components. Both the transmembrane domains and the periplasmic domains of FtsL and DivIC seem to be important for these interactions (Katis and Wake, 1999; Sievers and Errington, 2000; Robson et al., 2002). Thus, subcomplexes might preassemble before localization to the septum in B. subtilis as well as in E. coli.
In the yeast S. cerevisiae, cell division occurs by the formation of a bud (Guertin et al., 2002). The proteins required for division localize to the division plane in a sequential order. A group of proteins, septins, co-localize to the mother–bud neck and assemble into a ring that functions as a scaffold for other cell division components (Longtine et al., 1996; Frazier et al., 1998). A complex composed of three proteins exists, at least at the septum, in which myosin light-chain protein interacts independently with IQGAP protein and myosin class II protein (Boyne et al., 2000; Shannon and Li, 2000). The recruitment of IQGAP is dependent on myosin light chain, which in turn is dependent on the septin ring (Lippincott and Li, 1998). IQGAP also interacts directly with myosin class II protein, but this interaction is not required for localization (Boyne et al., 2000). Subsequently, actin and a GTPase are recruited to the actomyosin ring dependent on IQGAP to initiate constriction (Shannon and Li, 1999). Thus, as in E. coli, the assembly of proteins at the division site of this yeast appears to occur through sequential recruitment of proteins via direct interactions. Whether subcomplexes are assembled before localization to the division site in S. cerevisiae similar to the complex described in this paper is unknown. Further studies are necessary to determine just how similar E. coli and S. cerevisiae are in this regard.
Strains and media
Bacterial strains are listed in Table 1. Cells were grown in NZY medium (Guzman et al., 1992). When needed, ampicillin 200 µg ml−1, chloramphenicol 30 µg ml−1, kanamycin 40 µg ml−1 and spectinomycin 100 µg ml−1 were added. Medium was supplemented with 0.2%l-arabinose or 0.2%d-glucose to induce or repress the expression of genes controlled by the pBAD promoter respectively. IPTG (500 µg ml−1) was added to induce the expression of genes controlled by either the pTrc or the pLac promoter. Standard techniques, polymerase chain reaction (PCR), electroporation, transformation and P1 transduction were used for cloning and analysis of DNA (Miller, 1992). The enzymes for manipulating DNA were from New England Biolabs.
Table 1. Bacterial strains and plasmids.
Strain or plasmid
Relevant genetic marker(s) or feature(s)
Source or reference
F–ΔlacX74 galE galK thi rpsLΔphoA(ΔPvuII)
F–araD139ΔlacU169 relA1 rpsL150 thi mot flb-5301 deoC ptsF25 rbsR
For co-immunoprecipitation experiments and detection of FtsB by Western blotting, FtsB was tagged at the carboxy-terminus with a triple FLAG epitope. An oligo sequence encoding a triple FLAG tag was inserted directly between the XbaI and HindIII sites (underlined in sequence) in a pBAD18 plasmid, and the resulting plasmid is named pNB100. 3FLAGXH-5′ primer: 5′-CTAGAGACTACAAAGACCATGACG GTGATTAT AAAGATCATGACATCGATTACAAGGATGACGAT GACAAGTAGA-3′; 3FLAGXH-3′ primer: 5′-AGCTTCTACT TGTCA TCGTCATCCTTGTAATCGATGTCATGATCTTTA TAA TCACCGTCATGGTCTTTGTAGTCT-3′. ftsB was amplified by PCR using primers pBADfor (5′-AGATTAGCGGATCCTAC CTG-3′) and ftsBXbaIMYC (5′-CGCTCTAGATCGATTGTT TTGCCCCGCAGACTGTGC-3′). The stop codon of ftsB was replaced by an XbaI site in the ftsBXbaIMYC primer resulting in a translational fusion between ftsB and flag. The PCR product was digested with EcoRI and XbaI and ligated into the same sites of pNB100; the resulting plasmid was named pNB13. The Flag-tagged version of FtsB is designated ftsB-flag3. To construct pNB14, the ftsB-flag3EcoRI–HindIII fragment from pNB13 was inserted into a pBAD33 plasmid using the same restriction sites.
ftsL swaps were cloned under the control of a pLac promoter on low-copy plasmid pAM238. The various inserts were digested with EcoRI and HindIII from plasmids carrying the ftsL swap genes as described by Guzman et al. (1997) and Ghigo and Beckwith (2000) and ligated into pAM238 digested with EcoRI and HindIII. pNB15 contains LLL derived from pLD45, pNB16 contains FLL derived from pLD63, pNB17 contains LQL derived from pLD96, pNB18 contains LLhinfL derived from pJMG427, and pNB19 expresses LFL derived from pLD94.
For easy detection on Western blot after immunoprecipitation, ftsQ was tagged at the carboxy-terminus with a triple c-Myc tag. ftsQ was amplified by PCR using primers NWG0001 (5′-GGCTAGCGAATTCTGGAACT-3′) and NWG0002 (5′-CCCTCTGATTGTTGTTCTGCCTGTGCCTGAT-3′) and pLMG161 as template. The product was digested with EcoRI and XbaI and ligated into the same sites of pBAD18. Overlapping oligos (NWG0007: 5′-CTAGACTAGAACAAAAA CTCATCTCAGAAGAGGATCTGTA-3′ and NWG0008: 5′-AGCTTACAGATCCTCTTCTGAGATGAGTTTTTGTTCTAGT-3′) with compatible ends encoding a single c-Myc epitope were ligated in frame with ftsQ into the XbaI and HindIII sites, and the resulting plasmid was named pNG1. Because detection of FtsQ-Myc was weak, an extension with a double c-Myc was created. An annealed primer encoding a double c-Myc epitope (2MYCX-5: 5′-CTAGAGAACAGAAACT GATTTCTGAAGAAGA T CTGCTGGGTGAACAGAAACTGAT TTCTGAAGAAGATCTGCTGC-3′ and 2MYCX-3: 5′-CTAG GCAGCAGA T CTTCTTCAGAAA TCAGTTTCTGTTCACCCA GCAGATCTTCTTCAGAAATCAGTTTCTGTTCT-3′) was in-serted directly in the XbaI site resulting in plasmid pNB20.
A triple c-Myc epitope was also introduced at the C-terminal domain of FtsL. ftsL was amplified by PCR using pBADfor (5′-AGATTAGCGGATCCTACCTG-3′) and L-XbaI-Myc (5′-CGCTCTAGATTTTTGCACTACGATATTTTCTTGT GA-3′) as primers and pLMG180 as template DNA. The PCR product was digested with EcoRI and XbaI ligated into the same sites of pNG1. A double c-Myc epitope was then introduced in this plasmid as described above, resulting in plasmid pNB21.
As a positive control for FtsK depletion, a strain was constructed that expresses ftsK from a pBAD promoter on high-copy pBAD18-Km plasmid. The ftsK insert was obtained by digestion of pJC85 with SacI and XbaI and ligation into the same sites of pBAD18-Km resulting in plasmid pNB22.
Strains were grown overnight in NZ, diluted 1:100 in 50 ml of NZ and grown to an optical density at 600 nm (OD600) of 0.2. Expression of genes was induced for 1 h at 37°C by adding arabinose or glucose and IPTG to the cultures. Cells were harvested, and membranes were prepared as described below. Depletion strains were grown overnight in 5 ml of NZ in the presence of arabinose, diluted 1:100 in 5 ml of NZ with arabinose and grown to an OD600 of 0.5. Cells were pelleted by centrifugation at 5000 r.p.m. for 10 min and washed once with NZ without sugar to remove residual arabinose from the cells. The pellet was resuspended in NZ without sugar and diluted 1:50 in 50 ml of NZ containing arabinose or glucose and IPTG. The ftsL, ftsQ and ftsI depletion strains were grown for 3 h, and the ftsK depletion strain was grown for 4 h. These incubation times were determined experimentally such that cells showed extensive filamentation when grown in the presence of glucose, but no effect on growth, as measured by a drop in OD600, was observed indicating that cells were intact upon harvesting.
Preparation of membranes
Membrane fractions were prepared as described by Fraipont et al. (1994). Cells were harvested by centrifugation at 5000 r.p.m. for 15 min and washed once in distilled water. Cells were resuspended in 4 ml of buffer A (30 mM Tris-HCl, pH 8.0, 20% sucrose and 5 mM EDTA). Lysozyme was added to a final concentration of 0.1 mg ml−1 from a freshly made stock solution of 10 mg ml−1. Cells were incubated on ice for 30 min. Spheroplasts were stabilized by the addition of 15 mM CaCl2 and 150 mM NaCl and pelleted by centrifugation at 38 000 g in a Sorvall RC 5C centrifuge for 15 min. The pellet was resuspended in 2 ml of buffer B (10 mM Tris-HCl, pH 8.0, 150 mM NaCl and 10 mM MgCl2), and DNase was added to a final concentration of 20 µg ml−1. Spheroplasts were lysed by freeze–thawing steps. Membranes were pelleted by centrifugation at 100 000 g in a Beckman TL100 ultracentrifuge for 45 min. The membrane pellet was resuspended in 400 µl of buffer C (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5 mM EDTA, 1% Triton X-100) to solubilize membrane proteins. Undissolved material was removed by centrifugation at 100 000 g for 30 min. The supernatant contained solubilized membrane proteins and was used in co-immunoprecipitation experiments.
Immunoprecipitation experiments were similar to those described by Duong and Wickner (1997). Protein concentrations were determined by BCA assay (Pierce). A total of 100 µg of solubilized membrane protein fraction in a final volume of 100 µl was precipitated with 10 µl of preimmune serum, α-FtsL, α-FLAG (corresponding to 8 µg of antibody; obtained from Sigma) or α-FtsQ antibodies by incubation at 16°C overnight. IgGSorb (The Enzyme Center) was washed three times with buffer C, and 100 µl of solution was incubated with each immunoprecipitation sample for 30 min on ice. The IgGSorb–antibody–protein complexes were washed three times with cold buffer C to remove unbound material. The protein samples were incubated with 100 µl of sample buffer (New England Biolabs) containing 62.5 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 0.01% phenol red and 41.6 mM dithiothreitol (DTT) at 37°C for 15 min, and 10 µl of each sample was loaded per lane on SDS–PAGE.
Electrophoresis and immunoblotting were performed as described by Laemmli (1970) and Towbin et al. (1979) respectively. The nitrocellulose membranes were probed with α-FtsL polyclonal antibodies (1:5000 dilution) (Ghigo et al., 1999), with α-FtsQ monoclonal antibodies (1:1000) (Buddelmeijer et al., 1998), with α-FLAG monoclonal antibody (1:1000, Sigma), with α-FtsK polyclonal antibodies (a gift from J. Lutkenhaus) or with α-Myc polyclonal antibodies (1:1000, Sigma) and subsequently incubated with horseradish peroxidase-conjugated goat-anti-rabbit or sheep anti-mouse antibodies and developed with chemiluminescence reagents (Amersham Pharmacia).
We thank members of the Beckwith laboratory for assistance and encouragement. We thank J. Lutkenhaus for FtsK antibodies. Jon Beckwith is an American Cancer Society Professor. This work was supported by a grant from the National Institute of General Medical Sciences, GM38922.