In Escherichia coli the Min system prevents Z ring assembly at cell poles by topologically regulating the division inhibitor MinC. The MinC protein has two domains of equal size and both domains can target FtsZ and block cell division in the proper context. Recently, we have shown that, along with MinD, the C-terminal domain of MinC (MinCC) competes with FtsA, and to a lesser extent with ZipA, for interaction with the C-terminal tail of FtsZ to block division. Here we explored the interaction between the N-terminal domain of MinC (MinCN) and FtsZ. A search for mutations in ftsZ that confer resistance to MinCN identified an α-helix at the interface of FtsZ subunits as being critical for the activity of MinCN. Focusing on one such mutant FtsZ–N280D, we showed that it greatly reduced the FtsZ–MinC interaction and was resistant to MinCN both in vivo and in vitro. With these results, an updated model for the action of MinC on FtsZ is proposed: MinC interacts with FtsZ to disrupt two interactions, FtsZ–FtsA/ZipA and FtsZ–FtsZ, both of which are essential for Z ring formation.
Rod-shaped bacteria, such as Escherichia coli and B. subtilis, normally divide at the midpoint of the long axis of the cell to produce two equally sized daughter cells. How a cell finds its geometric centre to assemble the division machinery is a fundamental question in bacterial cell biology. So far, two negative regulatory systems, nucleoid occlusion (NOC) and Min, are known to be involved in the spatial control of cytokinesis (Lutkenhaus, 2007). These systems position division inhibitors within the cell in such a way that Z ring formation is restricted to the middle of the long axis of the cell. Although neither system is essential, inactivation of both is synthetic lethal due to an inability to assemble functional Z rings (Wu and Errington, 2004; Bernhardt and de Boer, 2005).
Nucleoid occlusion inhibits Z ring formation over the nucleoid and prevents guillotining of the chromosome by the cell division apparatus (Mulder and Woldringh, 1989; Yu and Margolin, 1999). In E. coli SlmA mediates NOC by interacting with both FtsZ and DNA, although the detailed mechanism is not known (Bernhardt and de Boer, 2005). In B. subtilis, a protein showing no homology to SlmA, called Noc, plays a similar role (Wu and Errington, 2004). Its cellular location is restricted through binding to specific DNA sequences that are scattered around the chromosome but absent from the terminus region. As the replicating chromosome segregates, a Noc-free space is generated around the terminus region at midcell allowing Z ring assembly (Wu et al., 2009).
In E. coli, the Min system consists of three proteins (MinC, MinD and MinE) that cooperate to inhibit Z ring assembly at cell poles (De boer et al., 1989). MinC is a division inhibitor, which is able to prevent Z ring formation and cause filamentation when overproduced even in the absence of MinD and MinE (Hu et al., 1999). MinD is an ATPase capable of binding and activating MinC by recruiting it to the membrane (Hu and Lutkenhaus, 2000; Hu et al., 2002; Lackner et al., 2003). The MinC/MinD complex is a potent division inhibitor that is evenly distributed on the membrane in the absence of MinE (De boer et al., 1989). In the presence of MinE the MinC/MinD complex is spatially regulated and its activity is directed to the poles of the cell. MinE does this by stimulating the pole to pole oscillation of MinC/MinD through its ability to stimulate the ATPase activity of MinD and thus, the release of MinD from the membrane (Hu and Lutkenhaus, 1999; Raskin and de Boer, 1999; Fu et al., 2001; Hale et al., 2001; Hu and Lutkenhaus, 2001; Hu et al., 2003). Oscillation of the MinC/MinD complex causes its time-averaged concentration to be highest at the cell poles and lowest at midcell, resulting in a narrow zone at midcell permissive for Z ring formation (Meinhardt and de Boer, 2001).
MinC has two functional domains of similar size (Hu and Lutkenhaus, 2000; Cordell et al., 2001). Both domains are required for the proper function of the Min system as mutations inactivating either domain inactivate Min as evidenced by minicell production (Zhou and Lutkenhaus, 2005). The two domains have been studied separately to elucidate their inhibitory mechanisms. The N-terminal domain [MinC1–115 (MinCN)] interacts with FtsZ and is able to block cell division when overexpressed, even in the absence of MinD (Hu and Lutkenhaus, 2000). In vitro MinCN prevents the sedimentation of FtsZ polymers. Since it does not inhibit the GTPase activity of FtsZ, it is thought to act after polymerization to shorten FtsZ polymers (Dajkovic et al., 2008a). The C-terminal domain of MinC (MinC116–231[MinCC]) mediates homodimerization and interaction with MinD (Hu and Lutkenhaus, 2000). In contrast to MinCN, MinCC does not prevent the sedimentation of FtsZ polymers nor affect their length in vitro, although it does reduce their lateral interactions (Dajkovic et al., 2008a,b). Even though by itself MinCC does not affect cell division, it can also target FtsZ and inhibit division in the presence of MinD (Johnson et al., 2002; Shiomi and Margolin, 2007; Shen and Lutkenhaus, 2009).
In a recent study we examined the effect of MinCC/MinD on Z ring formation and found that the inhibitory activity of MinCC/MinD requires the conserved C-terminal tail of FtsZ that is also involved in interaction with FtsA and ZipA (Shen and Lutkenhaus, 2009). FtsA and ZipA are essential because their interaction with FtsZ is required for Z ring formation and subsequent assembly of a complete divisome (Hale and de Boer, 1997; Haney et al., 2001; Pichoff and Lutkenhaus, 2002). MinCC/MinD antagonizes the function of the Z ring in a concentration-dependent manner. At low concentrations, it displaces FtsA from the Z ring so that downstream proteins are not recruited and the ring cannot constrict. At higher concentrations, it probably also displaces ZipA and completely disrupts the Z ring (Shen and Lutkenhaus, 2009). In this paper, we examined the interaction between MinCN and FtsZ through the isolation of FtsZ mutants that are specifically resistant to MinCN. A detailed study of one of these mutants indicates how MinCN interacts with FtsZ and leads us to propose a more detailed mechanism of how MinC antagonizes Z ring formation.
Mutations mapping to two regions of FtsZ confer resistance to MinC/MinD
Previously, we screened an FtsZ mutant library for resistance to MinCC/MinD and identified four such mutants, which altered the C-terminal tail of FtsZ (FtsZ – D373E, I374V, L378V and K380M) (Shen and Lutkenhaus, 2009). Since MinC has two functional domains that affect FtsZ differently (Hu and Lutkenhaus, 2000; Shiomi and Margolin, 2007), we screened the same FtsZ mutant library with full-length MinC/MinD, hoping to identify additional mutants, some of which might be resistant to MinCN. This approach is possible since resistance to one domain of MinC results in a loss of synergy between the two domains of MinC and therefore confers some level of resistance to MinC/MinD (Shen and Lutkenhaus, 2009). To do this, we introduced the plasmid (pBANG59/Ptac::minCD) expressing MinC/MinD under IPTG control into the strain (S7/W3110 ftsZ0 min::kan recA::Tn10) containing the mutagenzied ftsZ library and selected with 1 mM IPTG (cells with WT ftsZ are unable to form colonies at or above 0.1 mM IPTG). Survivors were isolated and mutations in ftsZ were identified.
Sequence analysis revealed that these MinC/MinD-resistant mutants contain mutations that primarily alter amino acids in two regions of FtsZ (Fig. 1A): the extreme C-terminus of FtsZ (including ftsZ–I374V and ftsZ–L378V, which were previously identified in the screen using MinCC/MinD) and an α-helix [H-10 helix (Oliva et al., 2004)] (ftsZ–R271G, ftsZ–E276D and ftsZ–N280S) that lies at the end of the FtsZ molecule opposite the GTP-binding site (Fig. 1B). Mutants in this latter group are likely to be resistant to MinCN, since the MinCC/MinD-resistant mutations map to the extreme C-terminus of FtsZ (Shen and Lutkenhaus, 2009). We also found another mutation (ftsZ–L205M) that conferred resistance to MinC/MinD, but it mapped to a third location on FtsZ. However, its resistance is probably due to a decreased GTPase of the mutant protein because this altered residue: (i) is located in the T7 loop of FtsZ (Fig. 1B, pink; Oliva et al., 2004), which is involved in GTP hydrolysis (Lowe and Amos, 1998); and (ii) is very close to the residue altered in the ftsZ2 mutant (FtsZ–D212G) that is known to be resistant to MinC/MinD and has reduced GTPase activity (Bi and Lutkenhaus, 1990; Dai et al., 1994). A reduction in the GTPase activity of FtsZ appears to be a common mechanism of resistance to division inhibitors, such as SulA and MinC (Dajkovic et al., 2008a,b). The reduced GTPase activity slows down polymer disassembly, which shifts the equilibrium to assembled polymers and reduces the sensitivity to these inhibitors. For this reason (and its MinC/MinD resistance is intermediate), the FtsZ–L205M mutant was not studied further.
Isolation of FtsZ mutants resistant to MinCN
Before studying any of the above potential MinCN-resistant mutants in detail, we performed another screen looking for MinCN-resistant mutants in a more direct way. We did a PCR random mutagenesis of the ftsZ–I374V allele and constructed an ftsZ–I374V mutant library using the same method as described before (Shen and Lutkenhaus, 2009). Since FtsZ–I374V shows some resistance to MinC/MinD (a strain carrying this mutation survives following induction of MinC/MinD from the plasmid pBANG59), the resultant library was screened using plasmid pBANG78/Plac::minCD, which prevents the ftsZ–I374V strain from forming colonies at IPTG ≥ 25 µM. This plasmid produces a higher level of MinC/MinD than the one (pBANG59) used above to screen the WT–ftsZ-based library.
Mutants surviving this selection are expected to have the ftsZ–I374V mutation that confers resistance to MinCC/MinD and an additional mutation(s) conferring resistance to MinCN. Surprisingly, using this approach we only obtained two mutants, both of which contained the ftsZ–I374V mutation and a mutation that altered residue N280 (ftsZ–N280D and ftsZ–N280T; both were obtained multiple times). A mutation altering residue N280 (ftsZ–N280S), as well as two additional mutations altering residues in helix H-10 (ftsZ–R271G and ftsZ–E276D), was obtained in the previous screen using the mutagenized ftsZ–WT allele and selecting with a lower level of MinC/MinD (Fig. 1A). Together, these results suggest that the H-10 helix of FtsZ is critical for the activity of MinCN. Notice that this α-helix lies at the interface of FtsZ subunits in the polymer (Fig. 1B and Oliva et al., 2004), which may be a clue in understanding how MinCN attacks FtsZ.
Characterization of the FtsZ–N280D mutant
The above analysis indicated the importance of the N280 residue of FtsZ in the MinCN–FtsZ interaction. Comparison of all the potential MinCN-resistant mutations (ftsZ–R271G, ftsZ–E276D and the various ftsZ–N280 mutations in the absence of the ftsZ–I374V mutation) indicated that ftsZ–N280D displays the most MinC/MinD resistance (data not shown) and was therefore chosen for subsequent studies. To assess the effect of this mutation on the cell phenotype, we first placed it onto the chromosome at the native ftsZ locus using the lambda RED recombineering system (Datsenko and Wanner, 2000) as described before (Shen and Lutkenhaus, 2009). We also recombined the allele containing the two mutations (ftsZ–I374V + ftsZ–N280D– referred to as ftsZ23) to the chromosome. The resultant strains were designated BSZ280D (ftsZ–N280D) and BSZ23 (ftsZ23). Western blot analysis of these strains revealed that FtsZ mutant proteins were as stable as wild-type FtsZ (Fig. S1).
The morphology of the BSZ280D strain was similar to the FtsZ–WT strain except that the average cell length of the BSZ280D strain is slightly longer due to the occasional long cells in the population (Table S1). It did not produce minicells. Inactivation of the Min system in BSZ280D resulted in minicell formation and caused mild filamentation [on average cells were two- to threefold longer compared with typical min- cells such as S4 (Table S1)] in early exponential phase but not in stationary phase. This is likely due to the FtsZ–N280D mutant having somewhat compromised FtsZ activity even though it is still able to complement an ftsZ0 strain and support division (see later). The mild filamentation was not obvious in the Min+ strain BSZ280D, indicating that the Min system is having a positive effect in this strain (the length of most of these cells is very close to WT cells; however, the presence of occasional very long cells results in a longer average cell length than the WT strain). One possibility is that by eliminating polar Z rings, the Min system makes more FtsZ molecules available for assembly of the midcell Z ring. Such activity may compensate for the compromised FtsZ activity caused by the ftsZ–N280D mutation. This scenario suggests that the FtsZ–N280D mutant is still responding to MinC/MinD to some extent, which is discussed later.
In contrast to BSZ374V [which was described previously (Shen and Lutkenhaus, 2009)] and BSZ280D, the BSZ23 strain produces many minicells [about 21% of the total constrictions are at the poles in an exponentially growing culture (Table S1)] and has the typical heterogeneous cell length distribution observed in min- strains. It also displays the mild filamentation phenotype in exponential growth phase that was observed with the BSM280D strain (Table S1). Inactivation of the Min system in the BSZ23 strain did not significantly change the morphology or minicelling phenotype except that it further increases the average cell length slightly (Table S1). Together, these results indicate that the presence of the ftsZ–I374V + ftsZ–N280D mutations in a strain completely suppresses the activity of the Min system.
FtsZ–N280D and FtsZ–I374V are resistant to the N and C terminal domains of MinC respectively
Having the various mutations (ftsZ–N280D, ftsZ–I374V and ftsZ–23) on the chromosome in a Δmin background allowed us to confirm their MinC/MinD resistance. An ftsZ–WT strain (S4/ftsZ–WT min::kan) containing the minC/minD low expression plasmid (pBANG59/Ptac::minCD) fails to form colonies at or above 50 µM IPTG (Fig. 2). However, the mutant strains (BSM280D/ftsZ–N280D min::kan; BSM374V/ftsZ–I374V min::kan; BSM23/ftsZ–23 min::kan) harbouring the same plasmid survive at 1 mM IPTG, indicating that all mutants have significant MinC/MinD resistance.
As discussed above, mutations (such as ftsZ2) that decrease the GTPase activity of FtsZ confer resistance to the division inhibitors MinC/MinD and SulA. As an initial test to determine whether the increased MinC/MinD resistance of these mutants might be due to reduced GTPase activity of these mutant proteins, we checked their SulA sensitivity. None of the mutant strains displayed increased resistance to SulA (Fig. S2). In fact, BSZ280D and BSZ23 were slightly more sensitive to SulA than the wild-type strain. This may be due to the ftsZ–N280D mutation compromising FtsZ activity. Nonetheless, these data suggest that the GTPase activity of these mutant proteins was not significantly reduced in vivo.
As described previously, FtsZ–I374V is resistant to MinCC but still sensitive to MinCN (Shen and Lutkenhaus, 2009). To see how FtsZ–N280D responds to the two domains of MinC, we used two minC mutations, minC–G10D and minC–R172A, which eliminate the toxicity of the N and C terminal domains of MinC respectively (Hu et al., 1999; Zhou and Lutkenhaus, 2005). As shown in Fig. 3, when the two MinC mutants (along with MinD) are expressed from a plasmid (pBANG78) under the control of an IPTG-inducible promoter, they prevent colony formation of the FtsZ–WT strain (S4/S3 min::kan) at about the same level (30 µM IPTG), indicating that the two domains of MinC have similar toxicity as reported previously (Shen and Lutkenhaus, 2009). When these MinC mutants are expressed in the FtsZ–N280D strain (BSM280D/BSZ280D min::kan), only MinC–G10D/MinD is toxic (Fig. 3). MinC–G10D/MinD prevents colony formation of this strain slightly more efficiently than in the FtsZ–WT strain, indicating that the FtsZ–N280D mutant is a little more susceptible to MinCC/MinD compared with FtsZ–WT. This was further confirmed using MinCC/MinD (Fig. S3) and is probably due to the reduced activity of the FtsZ–N280D protein. In contrast, expression of MinC–R172A/MinD was unable to prevent colony formation of the FtsZ–N280D mutant, indicating this FtsZ mutant is resistant to MinCN. Taken together, these results demonstrate that the FtsZ–N280D mutant is resistant to MinCN but still sensitive to MinCC. Also, the resistance of this mutant to the low level of MinC/MinD observed in Fig. 2 must be due to its resistance to MinCN. This behaviour is just the opposite of the FtsZ–I374V mutant, which is resistant to MinCC but sensitive to MinCN. As expected, neither domain of MinC is able to prevent colony formation in a strain containing ftsZ23 (Fig. 3). This allele also confers resistance to a high level of full-length MinC/MinD. The conclusion from these studies is that FtsZ–N280D and FtsZ–I374V are resistant to the N and C terminal domains of MinC, respectively, and that combining the two mutations renders cells completely resistant to MinC/MinD.
GFP–MinC/MinD localizes to the Z rings in the FtsZ–N280D mutant
GFP–MinCC/MinD was previously shown to localize to the Z ring and this was dependent upon the C-terminal tail of FtsZ (Johnson et al., 2004; Zhou and Lutkenhaus, 2005; Shen and Lutkenhaus, 2009). In contrast, localization of GFP–MinC/MinD to Z rings is difficult to observe (Johnson et al., 2002); it is very toxic and disrupts Z rings and causes filamentation before the fluorescent signal is observed. However, it should be possible to observe GFP–MinC/MinD at the Z ring in the FtsZ–N280D mutant because it displays significant resistance to MinC/MinD (due to resistance to MinCN), but does not affect the interaction between MinCC/MinD and FtsZ.
To confirm this, we introduced the plasmid pBANG85 (Ptrc::gfp–minCD) expressing GFP–MinC/MinD under IPTG control into the ftsZ0 strain S7 (W3110 ftsZ0min::kan recA::Tn10) containing derivatives of pBANG112 expressing various alleles of ftsZ, and examined the fluorescent signal at different IPTG concentrations. We used this approach since this plasmid (pBANG85) could not be introduced into the control strain S4 (S3 min::kan) due to the toxicity associated with the basal expression of GFP–MinC/MinD. In strain S7/pBANG112 FtsZ is produced from the plasmid and the level is slightly higher than the chromosomal level (about 1.5–2 fold, Fig. S1), which allows introduction of the pBANG85 plasmid.
As shown in Fig. 4A′, when GFP–MinC/MinD is expressed in the FtsZ–WT strain (S7/pBANG112), it causes filamentation even at a very low induction level (IPTG = 2.5 µM), where the fluorescence can barely be detected. The weak signal appears to be largely on the membrane. However, very occasionally weak fluorescent bands are observed suggestive of GFP–MinC/MinD association with Z rings in the process of being dismantled (arrow in Fig. 4A′). Induction of GFP–MinC/MinD at an intermediate level (IPTG = 7.5 µM) in the FtsZ–N280D mutant strain (S7/pBANG112–ftsZ N280D) results in strong fluorescent bands indicative of localization to Z rings (Fig. 4B′). When induced at a higher level, GFP–MinC/MinD disrupts the Z ring and causes filamentation and therefore fails to localize (data not shown). These results are consistent with FtsZ–N280D being resistant to MinCN but sensitive to MinCC/MinD. As a control, GFP–MinC/MinD was induced in the FtsZ–23 strain (S7/pBANG112–ftsZ23) at the same level as in the FtsZ–N280D mutant. As shown in Fig. 4C′, the flourescence is on the membrane but does not localize to Z rings (in a fraction of these cells the GFP also accumulates as spots along the cell membrane but the basis for this phenomenon is not known). This lack of localization is consistent with what we observed with the FtsZ–I374V mutant (Shen and Lutkenhaus, 2009). Together, these results confirm that the localization of MinC/MinD to the Z ring is dependent upon the MinCC/MinD–FtsZ interaction and that the ftsZ–N280D mutation is resistant to the action of MinCN.
FtsZ–N280D has reduced interaction with MinC and MinCNin vitro
The increased resistance of the FtsZ–N280D mutant to MinC/MinD and MinCNin vivo prompted us to check if the FtsZ–MinC and FtsZ–MinCN interactions are altered by this mutation in vitro. To this end, we did three tests.
First, we did a sedimentation assay to test the ability of MinC and MinCN to antagonize FtsZ polymer assembly. We purified the FtsZ–N280D and FtsZ23 proteins and tested the effect of MinC or MinCN in a sedimentation assay as described previously (Hu et al., 1999). In a preliminary sedimentation assay containing 5 µM FtsZ and 1 mM GTP (or GDP as control) in polymerization buffer, these proteins (FtsZ wild-type and mutants including FtsZ–I374V isolated previously) had very similar polymerization efficiencies with GTP and did not assemble with GDP (data not shown). We also checked the GTPase of these mutants using a NADH-coupled enzymatic assay (Chen and Erickson, 2009). The FtsZ–N280D mutant has decreased GTPase activity that is ∼60% of the wild-type protein (Fig. S4). The GTPase activity of FtsZ23 is similar to that of FtsZ–N280D whereas the GTPase of FtsZ–I374V is similar to the WT protein. By assaying the GTPase activity at various protein concentrations we determined that the FtsZ–N280D mutant has a modest assembly deficiency [the critical concentration for polymerization of this mutant is around 2.5 µM, which is about 1.5 µM higher than the WT protein (Fig. S4)]. This is perhaps not too surprising as this mutation alters a residue in the H-10 helix, which is at the interface between FtsZ subunits (Fig. 1B). Importantly, this mutant is slightly more sensitive to SulA than the FtsZ–WT in vivo and is as sensitive to MalE–SulA as FtsZ–WT in vitro (data not shown). These results indicate that MalE–SulA blocks FtsZ–N280D assembly as efficiently as FtsZ–WT assembly, which means that at this concentration (5 µM) FtsZ–N280D is turning over fast enough to respond to inhibitors such as SulA. Therefore, we used this concentration (5 µM) for the tests with MinC described below.
To test how the ftsZ–N280D mutation affects the interaction between FtsZ and MinC or MinCN, we did a sedimentation assay as described previously (Hu and Lutkenhaus, 2000), in which increasing amounts of MalE–MinCN were added to the FtsZ polymerization reaction. As shown in Fig. 5A, the amount of FtsZ–WT in the pellet decreases as the MalE–MinCN concentration increases, which is consistent with what was reported before (Hu and Lutkenhaus, 2000) and indicates that MinCN is blocking FtsZ sedimentation. However, when the FtsZ–N280D mutant protein was used in this test, the amount of FtsZ in the pellet was not affected by the amount of MalE–MinCN in the reaction (Fig. 5B). We also did this test with MalE–MinC; the result is the same as with MalE–MinCN– the sedimentation of FtsZ–WT but not FtsZ–N280D is inhibited by MalE–MinC in a concentration-dependent manner (Fig. S5). These results demonstrate that FtsZ–N280D has significant resistance to MinC and MinCNin vitro.
Second, we performed a far Western blot to examine the interaction between FtsZ and MinC or MinCN. For this test, MalE–MinC or MalE–MinCN was run on a native PAGE gel, transferred to a nitrocellulose membrane, which was then incubated with 5 µM FtsZ (WT or the N280D mutant), and the FtsZ bound to MalE–MinC or MalE–MinCN was detected using typical Western blot methodology. The results in Fig. 6 demonstrate that an interaction between MalE–MinC or MalE–MinCN and FtsZ–WT can be detected in this assay. However, the interaction with FtsZ–N280D is greatly reduced, demonstrating a decreased interaction between FtsZ–N280D and MinC or MinCN. As a control MalE–SulA was analysed since both FtsZ–WT and FtsZ–N280D show similar sensitivity to SulA in vivo (Fig. S2). The result shows that FtsZ–N280D binds MalE–SulA as efficiently as FtsZ–WT (Fig. 6). In addition, this result indicates that the decreased signal observed between MinC/MinCN andFtsZ–N280D is not due to poorer antibody detection of the FtsZ–N280D protein.
Third, we did a biosensor assay to examine the affinity between FtsZ and MinC. FtsZ–WT and FtsZ–N280D were biotinylated and immobilized to sensor chips containing covalently linked streptavidin. MalE–MinC at various concentrations was then injected and the response analysed. The calculated KD is about 6 µM for WT FtsZ. However, we were unable to calculate a KD for the FtsZ–N280D mutant because its affinity for MinC is beyond the detection range (10 pM–100 µM) of the system (Fig. S6). This means that the KD of FtsZ–N280D for MinC is above 100 µM, which is significantly higher than for FtsZ–WT, confirming a decreased interaction between FtsZ–N280D and MinC.
FtsZ–23 is synthetic lethal with SlmA
As mentioned above, the mutant strains BSZ280D and BSZ374V do not produce minicells even though they have some resistance to MinC/MinD. This suggests that these FtsZ mutants are still responding to MinC/MinD effectively so that topological regulation by the Min system is occurring (each mutant retains responsiveness to one of the two domains of MinC). However for the FtsZ–N280D mutant, the non-minicelling phenotype may not be that informative as it has reduced FtsZ activity [as evidenced by the increased critical concentration (Fig. S4)]. This reduced activity might counteract its MinC/MinD resistance in minicell production. To test this under conditions where FtsZ activity is not limiting we used an ftsZ0 strain (S18/W3110 ftsZ0 min+recA::Tn10) complemented with different ftsZ alleles present on a plasmid (pBANG112). As mentioned above, this plasmid expresses about 1.5- to 2-fold of the chromosomal level FtsZ (Fig. S1), and therefore could counteract the reduced FtsZ activity caused by mutations such as ftsZ–N280D. Consistent with the higher level of FtsZ provided by the plasmid, minicells are observed and about 5% of the total constrictions are polar when wild-type FtsZ is present on the plasmid (S18/pBANG112) (Fig. 7). With ftsZ–N280D the number of minicells increases and about 25–30% of the total constrictions are at the poles. These results indicate that the Min system is less effective in this mutant. In contrast, with ftsZ–I374V the fraction of polar divisions is similar to what is observed with the FtsZ–WT strain. With ftsZ–23, about 35% of constrictions are polar, similar to the Δmin strain. Together, these results suggest that MinCN may be more important than MinCC in preventing polar divisions.
As an alternative approach to examine the interaction of the ftsZ alleles and Min, we determined if they were synthetic lethal with loss of slmA. It is known that inactivation of slmA is lethal if cells do not have a functional Min system (Bernhardt and de Boer, 2005). If the ftsZ mutations disrupt the Min function to a significant extent they should be synthetic lethal with loss of slmA. This approach was facilitated by the observation that the min slmA double mutant is lethal at low temperature (≤ 30°C) but not at high temperature (42°C) (S. Du and J. Lutkenhaus, unpubl. data). Since it is known that increased FtsZ can suppress this lethality, we suspect that the FtsZ protein level or activity is increased at high temperature.
The slmA::cat allele was introduced into strains with different ftsZ alleles in the min+ or Δmin background by P1 phage-mediated transduction and transductants were selected with chloramphenicol at 42°C. The transductants were then restreaked at 30°C and their growth was monitored. As expected, none of the strains containing the slmA::cat allele (regardless of the ftsZ allele, either ftsZ–WT, ftsZ–N280D, ftsZ–I374V or ftsZ–23) was able to form isolated colonies at 30°C in the Δmin background (data not shown). In the min+ background, only the strain containing the ftsZ23 allele displayed synthetic lethalality with loss of slmA (Fig. 8). The failure of the FtsZ–23 mutant to grow without slmA even in the presence of Min further indicates that the Min system is ineffective in cells containing these two ftsZ mutations and is consistent with the minicelling phenotype conferred by the ftsZ23 allele. However, the growth of strains carrying either of the single ftsZ mutations (ftsZ–N280D or ftsZ–I374V) indicates that Min function is not totally absent in these strains. This finding is consistent with the non-minicelling phenotype of the BSZ280D and BSZ374V mutants.
Critical to our understanding of the spatial regulation of cytokinesis by the Min system is the mechanism of action of MinC, an inhibitor of Z ring formation (De boer et al., 1989; Hu et al., 1999). MinC has two structural domains (Hu and Lutkenhaus, 2000; Cordell et al., 2001), each of which can interact with FtsZ and block cell division, although the separated domains are much less active than the intact MinC (Shen and Lutkenhaus, 2009). The isolation of mutations in ftsZ in the current and previous studies allows discrimination of the interaction between FtsZ and the two domains of MinC. Residues in the extreme C-terminus of FtsZ (represented by I374) are critical for MinCC/MinD–FtsZ interaction (Shen and Lutkenhaus, 2009), whereas residues in the H-10 helix containing N280 are essential for MinCN–FtsZ interaction. By targeting different regions of FtsZ the two domains of MinC affect different aspects of Z ring formation to achieve synergy in disrupting Z rings.
Our findings that the two domains of MinC depend upon different regions of FtsZ suggest that they antagonize Z ring assembly by different mechanisms. Indeed, MinCN is able to disrupt FtsZ polymer assembly in vitro (prevent sedimentation), whereas MinCC cannot, although it decreases lateral interactions between FtsZ filaments (Hu and Lutkenhaus, 2000). Instead, MinCC, in the presence of MinD, competes with FtsA and to a lesser extent with ZipA for interaction with the C-terminal tail of FtsZ. This competition can destabilize and ultimately disrupt the Z ring (Shen and Lutkenhaus, 2009), since the FtsA–FtsZ and ZipA–FtsZ interactions are essential for the formation and functionality of the Z ring (Haney et al., 2001; Pichoff and Lutkenhaus, 2002). In this study, we focused on the MinCN–FtsZ interaction to increase our understanding of the mechanism by which MinCN antagonizes Z ring assembly.
MinC (or more precisely MinCN) prevents the pelleting of FtsZ polymers in sedimentation assays (Hu et al., 1999) without significantly affecting the GTPase activity of FtsZ (confirmed here using the NADH coupled enzymatic assay), suggesting that MinC does not affect FtsZ polymerization per se. FRET studies actually indicate that the amount of FtsZ in the polymer form is unaffected by MinC, although EM studies showed that MinC results in shorter polymers (Dajkovic et al., 2008a). Dajkovic et al. (2008a) showed that MinC reduces the mechanical stability of FtsZ polymer networks, which may explain why FtsZ polymers pellet less efficiently when MinC is present. This activity is largely due to MinCN since MinCC does not affect the pelleting of FtsZ polymers even though it also decreases the elasticity of FtsZ networks by preventing FtsZ polymer bundling (Hu and Lutkenhaus, 2000; Dajkovic et al., 2008a). It was postulated that MinCN weakens the longitudinal interaction between FtsZ subunits in the polymer and therefore causes loss of polymer stiffness and induces polymer shortening (Dajkovic et al., 2008a).
Our results indicate that residues located on one face of the H-10 helix that lies at the interface of two FtsZ subunits within an FtsZ polymer are critical for the FtsZ–MinCN interaction. The location of these residues is consistent with the above idea that MinCN weakens the longitudinal interaction between FtsZ subunits. The FtsZ–N280D mutant was studied in detail. This protein has slightly decreased FtsZ activity in vivo and a small deficiency in the GTPase activity in vitro due to weaker interaction between FtsZ subunits as evidenced by a small increase in the critical concentration for polymerization. This is consistent with the involvement of the H-10 helix in the interaction between FtsZ subunits (Oliva et al., 2004). Nonetheless, FtsZ–N280D is resistant to MinCNin vivo and the sedimentation of FtsZ–N280D polymers is insensitive to MinCNin vitro. The resistance to MinCN is specific since FtsZ–N280D displays slightly increased sensitivity to SulA and MinCC/MinD. The reduced interaction between FtsZ–N280D and MinC/MinCN provides a basis for its MinCN resistance [notice that the affinity (KD) of the FtsZ–MinC interaction was about 6 µM in this study, which is different from what was reported before (≈ 1 µM (Hu et al., 1999)); this difference may be due to the use of two different systems to assess the interaction]. One interpretation of our data is that MinCN attacks the FtsZ dimer interface at the H-10 helix to break FtsZ polymers.
The H-10 helix is very close to the SulA binding site on FtsZ [represented by the F268 residue (Bi and Lutkenhaus, 1990; Dajkovic et al., 2008b)]. However, the mechanisms by which MinCN and SulA inhibit FtsZ ring formation are fundamentally different. SulA blocks the FtsZ GTPase and inhibits FtsZ polymerization, whereas MinCN does not (Mukherjee et al., 1998; Hu et al., 1999; Dajkovic et al., 2008a,b). Instead, MinCN causes FtsZ polymer shortening but the basis for this activity was not clear. The results obtained from this study prompt us to propose a model for the action of MinCN based upon the following observations: (i) MinCN shortens FtsZ polymers but does not affect the GTPase or polymerization of FtsZ per se (Dajkovic et al., 2008a), (ii) the GTPase of FtsZ is actually required for the inhibitory activity of MinCN (Dajkovic et al., 2008a), (iii) FtsZ polymers contain a significant amount of subunits in the GDP form [up to 50% at [GTP] > 100 µM) (Chen and Erickson, 2009)] and (iv) the ftsZ–N280D mutation decreases the interaction between FtsZ (GDP form) and MinC.
In our model MinC binds to polymerized FtsZ through the MinCC/MinD interaction with the conserved tail of FtsZ regardless of the nucleotide bound to FtsZ. This binding brings MinCN in close proximity to the FtsZ polymer, although the H-10 helix may or may not be fully accessible to MinCN (see Fig. 9). If GTP is at the FtsZ dimer interface (Fig. 9, No. 2), the strong FtsZ–FtsZ interaction retains the H-10 helix at the FtsZ dimer interface so that it is less available for the FtsZ–MinCN interaction. However, if GDP is present (Fig. 9, No. 1), the H-10 helix becomes available for MinCN binding and the polymer is then severed by MinCN. This process may be aided by the curvature of FtsZ filaments or thermal fluctuations. After breaking the FtsZ polymer, MinCN (or MinC) binding does not affect the rate of the FtsZ subunit release from the polymer (if it is in the GDP form, Fig. 9, No. 3) or the GTP hydrolysis rate (if it is in the GTP form, Fig. 9, No. 4). In this way, MinC/MinCN attacks and shortens the FtsZ polymers without significantly affecting the GTPase.
To test the possibility that the FtsZ interface containing GDP is the preferred target for MinCN, we generated GDP containing polymers by using DEAE-dextran (Mukherjee and Lutkenhaus, 1994; Trusca et al., 1998) and checked to see how MinC or MinCN affects these polymers in a sedimentation assay. We did not observe co-sedimentation of MinC/MinCN with these polymers nor did we detect a decrease in the amount of FtsZ polymer. However, this test is not conclusive since it is possible that the DEAE-dextran may coat the FtsZ polymer and block the interaction with MinC/MinCN. More sophisticated strategies are required to test this possibility. Alternatively, an approach that is closer to the physiological situation where both FtsZ polymers and MinCN (as part of MinC/MinD) are on the membrane may be necessary to examine the effect of MinCN on FtsZ.
The BSZ23 strain, containing both ftsZ mutations (ftsZ–I374V and ftsZ–N280D), produces minicells and behaves essentially like a min- strain, indicating that the Min system is totally ineffective. A puzzling observation is that the FtsZ single mutants (BSZ374V and BSZ280D) do not produce minicells even though they display some resistance to MinC/MinD. One possible explanation for the non-minicelling phenotype of the BSZ374V and BSZ280D strains is that their MinC/MinD resistance is insufficient (clearly not as great as BSZ23). If so, this would indicate that resistance to just one domain of MinC is not sufficient to produce minicells. However, mutations in MinC that inactivate either domain (MinC–G10D or MinC–R172A) cause minicell production (Zhou and Lutkenhaus, 2005 and unpublished data from M. Wissel and J. Lutkenhaus).
When we compare the effect of minC (for example minC–R172A) and ftsZ mutations (such as ftsZ–I374V) on the responsiveness of FtsZ to MinC in a colony forming assay, they are very similar (Fig. 3). Yet minC mutations cause minicell production but ftsZ mutations do not. Therefore, the degree of MinC/MinD resistance may not be able to explain everything. For the BSZ280D strain, the non-minicelling phenotype may in part be due to the decreased FtsZ activity compromising the MinC/MinD resistance. When we complement an ftsZ0 strain with the ftsZ–N280D allele from a plasmid (which makes slightly more than chromosomal level FtsZ), it leads to significant minicell production (Fig. 7). In contrast, the same strain complemented with ftsZ–I374V does not produce more minicells than the strain with ftsZ–WT. These results suggest that MinCN plays a more important role than MinCC in preventing polar Z ring formation. On the other hand, the MinC/MinD resistance of the ftsZ–I374V allele is at least similar (if not greater than) to the ftsZ–N280D allele (Fig. 3, the plasmid pBANG78 expressing wild-type MinC/MinD can be introduced into the BSM374V strain but not the BSM280D strain because the basal level of MinC/MinD is too toxic). However, the FtsZ–N280D strain makes minicells if the FtsZ activity is not a limiting factor, whereas FtsZ–I374V does not. This is another case where the extent of MinC/MinD resistance does not correlate with the minicelling phenotype. Therefore, we believe that there must be something we do not understand regarding the spatial regulation of cell division by the Min system, which deserves further investigation.
Strains, plasmids and growth conditions
Strains and plasmids used in this study are listed in Table 1. Cells were grown in Luria–Bertani (LB) medium at 37°C unless otherwise indicated. Antibiotics were used at the following concentrations as necessary: ampicillin = 100 µg ml−1; spectinomycin = 25 µg ml−1; kanamycin = 25 µg ml−1, tetracycline = 10 µg ml−1, and chloramphenicol = 10 µg ml−1.
The strain S18 (W3110 ftsZ0, recA::Tn10) was constructed in the same way as the S7 strain, which was described before (Shen and Lutkenhaus, 2009). The only difference is that the starting strain for S18 construction is S3 (which is min+) instead of S4 (which is min-). BSZ280D and BSZ23 were made by recombineering as previously described for the BSZ374V strain construction (Shen and Lutkenhaus, 2009). The slmA mutants were made by P1 phage-mediated transduction. The slmA::cat (with most of the slmA coding sequence replaced by the chloramphenicol-resistant gene cat) construct from the strain W3110 slmA::cat (from S. Pichoff) was transduced into strains S3, BSZ374V, BSZ280D, BSZ23, S4, BSM374V, BSM280D and BSM23 by P1 transduction and selection of transductants at 42°C to give the slmA knockout of corresponding strains.
The plasmid pBANG85 was constructed in two steps: first the tac promoter of the plasmid pBANG59 (Shen and Lutkenhaus, 2009) was replaced by the trc promoter from pDSW210 to give the plasmid pBANG84. The Ptrc+lacIq region from pDSW210 (Weiss et al., 1999) was PCR amplified (using GCAGATCTACGATGTCGCAGAGTATGCC and GAATTGGGACAACTCCAGTG as primers) and cloned into pBANG59 digested with BglII+EcoRI. Second, the region containing gfp–minCD from plasmid pMCW71 (M. Wissel and J. Lutkenhaus, unpublished) was cloned into pBANG84 by SstI+HindIII digestion and ligation.
PCR random mutagenesis and FtsZ mutants screening
The experimental details were described previously (Shen and Lutkenhaus, 2009). For each mutant library construction (ftsZ–WT based and ftsZ–I374V based), four independent PCR random mutagenesis reactions were performed and subsequent cloning resulted in four separate sub-libraries (each contains about 5000 colonies), each of these sub-libraries was screened using either the plasmid pBANG59 (for ftsZ–WT-based libraries) or pBANG78 (for ftsZ–I374V-based libraries). Subsequent sequencing analysis on those survivors indicated that the each ftsZ–WT-based sub-library usually contains 2–4 different MinC/MinD-resistant mutants (same mutations can be found in different sub-libraries), whereas every ftsZ–I374V-based sub-library has only one MinC/MinD-resistant mutant (two of the four sub-libraries contain the FtsZ–N280D mutant and the other two have the FtsZ–N280T mutant).
Analysis of the localization of GFP–MinC/MinD
Overnight cultures of S7/pBANG112 (ftsZ0min::kan/ftsZ–WT), S7/pBANG112–280D (ftsZ0min::kan/ftsZ–N280D) and S7/pBANG112-23 (ftsZ0min::kan/ftsZ–23) containing the plasmid pBANG85 (Ptrc::gfp–minCD) were diluted 1000-fold into LB containing spectinomycin and ampicllin and grown at 37°C until OD600 ≈ 0.05. IPTG was then added at the indicated concentrations and the cultures were grown for another 2–3 h to reach OD600 ≈ 0.4. Samples were taken and analysed by microscopy as previously described (Shen and Lutkenhaus, 2009).
Protein purification and FtsZ sedimantation assay
FtsZ–N280D and FtsZ–23 were expressed and purified from BSZ280D/pKD126-280D (ftsZ–N280D) and BSZ23/pKD126-23(ftsZ–23), respectively, using the method described previously (Shen and Lutkenhaus, 2009). The purification of MalE–MinCN/MalE–MinC and the FtsZ sedimentation assay was also done following the previously reported strategy (Hu et al., 1999). Reactions containing 5 µM FtsZ and the indicated amounts of MalE–MinCN were initiated by the addition of 1 mM GTP (or GDP as a control) and incubated for 5 min at room temperature. The samples were then centrifuged for 15 min × 80 000 r.p.m. at 25°C in the Beckman TLA 100.2 rotor, the pellets dissolved in SDS sample buffer and analysed by SDS-PAGE. We observed that at high concentrations of MalE–MinCN in the reactions containing FtsZ–N280D more MalE–MinCN appeared in the pellet than expected for background levels (indicated by ‘*’ in Fig. 5B). This is probably due to the non-specific association of MalE–MinCN with FtsZ in the pellet. When 16 µM heat-inactivated FtsZ–N280D was used in the sedimentation assay (heat inactivation was confirmed by the failure to respond to GTP in the sedimentation assay), we observed the same amount of FtsZ–N280D in the pellet as with reactions containing active FtsZ–N280D (5 µM) and GTP. We assume that the FtsZ–N280D in the pellet following heat inactivation is due to non-specific aggregation. If MalE–MinCN at various concentrations was present in these reactions we observed the same amount of MalE–MinCN in the pellets (using either 16 µM heat-inactivated FtsZ–N280D with GDP or 5 µM active FtsZ–N280D with GTP), indicating that it was non-specifically associated with the pelleted FtsZ.
MalE–MinC or MalE–MinCN (or MalE–SulA as control) was run on a 7.5% native PAGE gel (Bio-Rad, Cat. No. 161-1100) following the instructions coming with the gel. One microgram of protein was loaded into each of the 10 wells. The protein was then transferred to the nitrocellulose membrane following the standard Western blot protocol (the transfer buffer is the same as the PAGE running buffer). The membrane was then cut into three equal pieces. One piece was used for Ponceau-S staining to see the amount and position of the protein on the membrane. The other two pieces were first blocked with 2.5% milk in the FtsZ polymerization buffer (50 mM MES, 50 mM KCl, 10 mM MgCl2, pH = 6.5) for 1 h and then incubated in the same buffer (with milk) containing 5 µM FtsZ (WT or the FtsZ–N280D mutant, one piece of membrane for each) for about 1 h at room temperature with gentle shaking. After 5 min each × 3 washes with the FtsZ polymerization buffer, the membrane was blotted with FtsZ antiserum and detected with an AP-conjugated secondary antibody as in regular Western blots.
The instrument we used is the Biacore X and the sensor chips are the SA chips from GE Healthcare. FtsZ (WT or the FtsZ–N280D mutant) was biotinylated using EZ-Link Sulfo-NHS-LC-Biotinylation kit to the level of 1:1 (1 biotin to 1 FtsZ molecule) following the protocol provided by the manufacturer (Pierce, Cat. No. 21435). The biotinylated FtsZ was then immobilized onto the sensor chip by interacting with streptavidin, which was covalently linked to the chip surface. After the sensorgram reached a stabilized signal, MalE–MinC at various concentrations (1, 2, 4, 8, 16, 32 µM) was injected into the sensor flow cells at the rate of 10 µl min−1. The response was recorded and analysed by the BIAevaluation software. The buffer used was the HBS-P buffer from GE Healthcare and MalE–MinC dilutions were made in this buffer. After each MalE–MinC injection, the sensor chip was regenerated by HBS-P buffer containing 4 M NaCl.
We thank Dr D.J. Black at UMKC for help in doing the biosensor assay. This work was supported by National Institutes of Health Grant GM029764.