Cell division proteins FtsZ (FtsA, ZipA, ZapA), FtsE/X, FtsK, FtsQ, FtsL/B, FtsW, PBP3, FtsN and AmiC localize at mid cell in Escherichia coli in an interdependent order as listed. To investigate whether this reflects a time dependent maturation of the divisome, the average cell age at which FtsZ, FtsQ, FtsW, PBP3 and FtsN arrive at their destination was determined by immuno- and GFP-fluorescence microscopy of steady state grown cells at a variety of growth rates. Consistently, a time delay of 14–21 min, depending on the growth rate, between Z-ring formation and the mid cell recruitment of proteins down stream of FtsK was found. We suggest a two-step model for bacterial division in which the Z-ring is involved in the switch from cylindrical to polar peptidoglycan synthesis, whereas the much later localizing cell division proteins are responsible for the modification of the envelope shape into that of two new poles.
The FtsZ-ring with all the associated proteins is termed ‘divisome’. The interdependency of the localization of the cell division proteins suggests that each protein is added sequentially to the maturating divisome. In other words, this would reflect a sequence in time. However, not much is known about the time required for the assemblage of the divisome. We found that the FtsZ-ring, FtsA (Den Blaauwen et al., 1999) and ZapA (Den Blaauwen et al. unpubl. results) assemble about 20 min before the cell constricts. Do these 20 min reflect the time required to assemble the proteins that are recruited after Z-ring formation (i.e. FtsXE up to AmiC) or is the divisome assembled simultaneously with the Z-ring? Let us assume for the sake of argument that the divisomal protein complex can be compared with other large protein complexes. For instance, with the ribosomal 30S subunit that consists of 21 protein subunits and one 16S RNA molecule. This complex assembles in 2.5 min in cell cultures grown with a doubling time of 85 min at 28°C (Lindahl, 1975). A divisome maturation time of 20 min would then be exceptionally long. With only 25–100 mol of the majority of the cell division proteins per cell, a high affinity for the divisome seems to be a requirement for complex assembly. Unless the cell has to wait for the proteins to be synthesized, a high affinity also argues against a very time consuming assembly.
To answer the question, whether the divisome matures instantaneously or slowly, we have studied the divisome maturation time by determining the cell age at which the FtsZ-ring assembles and the age at which intermediate proteins in the interdependency localization sequence, like FtsQ and FtsW, and later in the dependency sequence recruited proteins, like FtsI (PBP3) and FtsN, assemble at mid cell.
Our results suggest that the assembly of the FtsZ-ring and its association with FtsA, ZipA and ZapA is separated in time from the assembly of the proteins FtsQ up to FtsN.
Timing of the localization of cell division proteins
To investigate whether the order in which cell division proteins localize reflects the timing of their localization, we studied the time difference between localization of FtsZ that initiates divisome maturation and of PBP3 and FtsN that are almost the last cell division proteins in the sequence pathway of divisome maturation.
To be able to correlate the localization of cell division proteins with each other and with cellular processes like DNA replication, nucleoid segregation and cell constriction, care should be taken to study these processes under comparable physiological conditions. For cell cycle parameters this requires steady state growth (Fishov et al., 1995). During steady state growth by definition the total cell mass increases at the same rate as the total cell number increases. Therefore, the average mass of all individual cells in the culture is constant. The advantage of steady state is that the cell metabolism and the enzymatic reaction rates are constant. A consequence is that the cells in the culture will have a constant age distribution and the relative frequency of cells in a certain age class will also remain constant, despite the fact that the absolute cell number in the population increases. Thus, the fraction F(x) of cells with a particular characteristic, that occurs at the end of the cell cycle (e.g. constricting, Z-ring containing or FtsN localizing) contains all cell ages at which the cells have this characteristic. By determining the fraction of cells with the aforementioned characteristics, the age at which these characteristics appears in the cell cycle can be calculated using Eq. 1 (see Experimental procedures).
The localization of cell division proteins to the site of division was studied by fluorescence microscopy. The immunolabelling procedure was shown to have no effect on the length distribution of the cells in the culture (Den Blaauwen et al., 1999).
The wild-type strain LMC500 (Table 1) was grown to steady state in minimal glucose medium (GB1) at 28°C, harvested, fixed and labelled with either a monoclonal antibody (mAb) against FtsZ, a polyclonal antiserum (pAb) against PBP3, or a pAb against FtsN and secondary antibodies conjugated to fluorophores as described in Experimental procedures. The cells were imaged by fluorescence microscopy to determine the number of cells with mid cell localization of FtsZ, PBP3, or FtsN (see Fig. 2 for an example of FtsN, PBP3 and FtsZ immunolabelling). Length and diameter of the cells were measured and the position of the fluorescent signal and the presence of a constriction were determined (Table 2).
Table 2. Morphological parameters of E. coli LMC500 and thereof derived strains and LMC500 expressing GFP-FtsN (pDML2002), GFP-FtsN (pDML2002), FtsN (pDML2000 and pDML2001), GFP-FtsW (pDML2414), or GFP-FtsQ (pTHV039) growing at 28°C.
The mass doubling (Td) time of LMC500 was 85 min The fraction of cells in the culture that contained a Z-ring, a PBP3, or an FtsN signal at mid cell was 0.53, 0.33 and 0.33 respectively. Using Eq. 1 and [(Td − tx)/Td] × 100 it can be calculated that the FtsZ-ring appears after 39% of the cell cycle of LMC500, and that PBP3 and FtsN localize at mid cell after 59% of the cell cycle (Table 2 and Fig. 3). Constrictions were visible after 62% of the cell cycle. Because the localization of cell division proteins is interdependent, the recruitment of all divisome proteins ranging from FtsZ up to FtsN seems to require approximately 17 min under the growth conditions used.
FtsQ, FtsW, PBP3 and FtsN arrive simultaneously at mid cell
To find out whether intermediate cell division proteins in the queue of the assembly line would arrive at mid cell at an intermediate time between that of FtsZ and FtsN, we studied the arrival time of FtsQ and FtsW. Although the localization of FtsQ has been demonstrated by immunolabelling of endogenous FtsQ by monoclonal antibodies (Buddelmeijer et al., 1998), the signal was too weak to be used to determine the timing of FtsQ localization. Therefore, an alternative approach was developed. First it was determined whether the temperature sensitive ftsQ1 allele in LMC531 (Taschner et al., 1988) that expresses an FtsQ mutant in which glutamic acid (E)125 is replaced by a lysine (K) (Storts and Markovitz, 1991) could be used as a conditional FtsQ deletion strain to determine the timing of GFP-FtsQ localization. We constructed a GFP fusion to the original mutant (GFP-FtsQE125K) to assess whether this protein was not functional at the restrictive temperature either because it was not able to localize or because downstream proteins were not able to localize or because the protein is very unstable. The GFP-FtsQE125K protein was expressed in LMC531(ftsQ1) from plasmid pTHV075. GFP-FtsQ wt expressed from the same vector (plasmid pTHV039) without induction, overproduces 70-fold compared with the endogenous FtsQ protein (Fig. 4). The number of FtsQ molecules produced was determined by immuno-chemiluminescence densitometry using an FtsQ monoclonal antibody (Fig. 4). The cells were grown to steady state in GB1 medium at 28°C and for 2 Tds at 42°C. GFP-FtsQE125K localized weakly at mid cell at the permissive temperature and not at all at the restrictive temperature (results not shown). Also no clear membrane localization was observed at the restrictive temperature. Therefore, it can be concluded that the endogenous FtsQE125K is not functional at the restrictive temperature because it is not able to localize presumably because it is very unstable.
Subsequently, a wt gfp-ftsQ gene fusion was made and inserted in the lambda attachment site at 17 min in the chromosome of LMC531 using the lambda Inch system (Boyd et al., 2000) resulting in the merodiploid strain LMC2328 (Table 1). This strain was grown to steady state in GB1 medium at 28°C and for 2 Tds at 42°C. It was determined by immuno-luminescence densitometry that the endogenous GFP-FtsQ protein was produced in fivefold excess compared with the endogenous FtsQ protein (Fig. 4). The cells showed normal morphology at 28°C (Table 2) and 42°C (results not shown). Because FtsQE125K localized weakly, it was assumed that the fivefold overproduced GFP-FtsQ would replace almost all FtsQE125K molecules in the divisome. Subsequently, the fraction of cells with a GFP-FtsQ signal at mid cell was determined in the steady state grown culture at 28°C in LMC2328. GFP-FtsQ appeared after 48% of the cell cycle (Table 2 and Fig. 3). FtsN and PBP3 appeared in LMC500 after 59% of the cell cycle under the same conditions (Table 2).
FtsW is dependent on the localization of the FtsQLB complex (Buddelmeijer and Beckwith, 2004) and is therefore also an intermediate protein between FtsQ and FtsN in the assembly pathway. We did not have antibodies against FtsW, therefore we used a GFP-FtsW fusion expressed from plasmid pDML2414 (Table 1) to determine its localization timing in the LMC500 background. After growing the cells to steady state, one part of the sample was directly used to determine the number of cells with an FtsW signal at mid cell and the other part of the culture was used to label FtsZ with monoclonal antibody against FtsZ. In the cells no effect was visible of the about fivefold overproduction of FtsW (Pastoret et al., 2004) and they had wild-type geometrical parameters. This GFP-FtsW fusion expressed from plasmid pDML2414 fully complemented the chromosomal encoded FtsW(TS) from E. coli JLB17 grown under the same conditions as in our localization experiments described above (Pastoret et al., 2004). FtsZ localized at 31% of the cell cycle and GFP-FtsW at 48% of the cell cycle (Fig. 5 and Table 2). A considerable period passed (17% of the cell cycle) between the localization of FtsZ and GFP-FtsW (Fig. 3). In conclusion, the cell age at which FtsQ, FtsW, PBP3 and FtsN localize at mid cell suggests that all proteins downstream of FtsK assemble more or less at the same time, which is just before the onset of constriction (Table 2) of the bacterial envelope.
Divisome maturation related to growth rate
At a Td time of 85 min a delay of approximately 16 min is observed between assembly of the Z-ring and the localization of the proteins FtsQ up to FtsN. What is the function of this delay? It seems not very likely that this time is required for the synthesis of the divisome proteins because at least PBP3 was shown to be present at a constant level during the cell cycle (Wientjes et al., 1983) and overproduction of FtsQ or FtsN does not decrease the time delay (Table 2 and Fig. 3). However, if these proteins assemble into a complex, the limited synthesis of one of the proteins would be sufficient for a delay in assembly of the whole complex. Diffusion of a protein through the membrane (Saffman and Delbruck, 1975) from one pole to the opposite pole takes about 8 s assuming a diffusion rate of 0.3 µm2 s−1 (Fujiwara et al., 2002). Therefore diffusion cannot be a localization limiting factor. If the delay in divisome maturation would reflect some kind of synthesis, the delay would be expected to be growth rate dependent. Therefore, the timing of the divisome maturation was determined at two other Td times. LMC500 cells were grown to steady state in rich Trypton/Yeast (TY) and minimal Ala-Pro (AP) medium at 28°C, which resulted in Td times of 40 and 140 min respectively (Table 2 and Fig. 3). The time between Z-ring formation and FtsN mid cell arrival was 14 and 21 min, respectively, or 34% and 15% of the cell cycle time. The time between Z-ring formation and FtsN mid cell arrival was 17 min or 20% of the cell cycle time in the GB1 grown LMC500 cells. Thus, from fast to slow growth (40–140 min) the amount of time used for divisome maturation increased from 14 to 21 min Assuming an exponential increase of the surface of a bacterium (Woldringh et al., 1985), the amount of cell surface synthesized during the Z-ring period can be calculated. At a Td of 140 min, 85 min and 40 min the total cell surface synthesis during the period in which the Z-ring is present is approximately 1.80 µm2, 2.54 µm2 and 9.0 µm2 respectively. This is similar to the surface of 1.96 µm2 and 2.55 µm2 of new cell pole material that has to be synthesized at the slow growth rates of 140 min and 85 min respectively. This suggests a relation between the period in which the Z-ring is present and the time needed for the synthesis of the surface of the new cell poles. The area of new cell poles to be synthesized at a growth rate of 40 min is 4 µm2, which is half of the surface synthesized during the Z-period. At fast growth rates the Z-ring appears almost simultaneously with birth and the envelope synthesis during the Z-ring period is therefore composed of cylindrical growth as well as cell pole synthesis. At slow growth the contribution of the synthesis of the cylindrical part is much less.
Timing of divisome maturation in cell division mutants
The cell division mutant ftsI2158 that expresses a temperature sensitive (TS) PBP3 G191D, D266N protein (Goffin et al., 1996) was reported to have a delayed DNA segregation period at the permissive temperature, whereas ftsZ84 that expresses an FtsZG105S(TS) protein (Bi and Lutkenhaus, 1990) and ftsQ1 TS mutants have normal DNA segregation periods (Huls et al., 1999). To determine whether the PBP3(TS) mutant showed a delay in DNA segregation as a result of a delay in divisome maturation time, we determined the timing of the localization of cell division proteins in LMC509 harbouring ftsZ84, LMC531 harbouring ftsQ1 that expresses FtsQ(TS) and LMC510 harbouring the ftsI2158 that expresses PBP3(TS) (Table 1). The cells were grown to steady state in 1/2 GB1 for LMC509 because the ftsZ84 mutation is salt sensitive and in GB1 for LMC531 and LMC510 at 28°C. The cells were harvested, fixed and labelled with antibodies against FtsZ or FtsN as described (Experimental procedures). The divisome maturation time was 17 min, 21 min and 16 min in LMC509, LMC531 and LMC510 respectively. Notably, LMC510 used 55% of its cell cycle to constrict (an activity that occurs after divisome maturation), whereas LMC531, LMC509 and wild-type LMC500 cells used 36%, 38% and 21% respectively (C in Table 2). These results show that there is no correlation between the DNA segregation delay observed in the PBP3 mutant (Huls et al., 1999) and the divisome maturation time as similar values were observed for LMC500, LMC509 and LMC510. In contrast, the LMC531 strain which showed the largest delay in divisome maturation time (Table 2) did not have a delay in DNA segregation time (Huls et al., 1999). It seems more likely that a prolonged constriction time in LMC510 caused the delay in DNA segregation because the other mutants and the wild type have a much shorter constriction time. In conclusion, no effect of divisome maturation time was observed on DNA segregation time. The results rather indicate that the constriction process assists the segregation of the nucleoids.
FtsN overproduction does not affect the divisome maturation time
FtsN was found as a suppressor of temperature sensitive cell division mutants (Dai et al., 1993). Because FtsN might stabilize the divisomal complex, we investigated whether a change in FtsN protein concentration would change the timing of the divisome maturation. LMC500 cells were transformed with plasmid pDML2499 (Table 1) or plasmid pDML2002 (Table 1) that expressed GFP-FtsN constitutively at a fourfold level compared with the expression of endogenous FtsN (see Experimental procedures on the quantification of FtsN and Fig. 6 for an example of an immunoblot developed with the antiserum against FtsN). Endogenous FtsN was determined to be present in about 1000 mol per average cell. Apparently, FtsN is about 10 times more abundant in an average cell than other cell division proteins such as FtsQ (Carson et al., 1991) and PBP3 (Dougherty et al., 1996).
The cells were grown to steady state in GB1 at 28°C, harvested and fixed as described in Experimental procedures. One part of the cells was directly imaged by fluorescence microscopy to determine the localization of GFP-FtsN (Fig. 5), a second part was imaged after labelling of FtsZ, or PBP3 as described above. Length and diameter of the cells were measured and the position of the fluorescent signal and the presence of a constriction were determined (Table 2).
The Td time of LMC500 transformed with the GFP-FtsN expressing plasmids was 93 min The fraction of cells with an FtsZ-ring, PBP3 or GFP-FtsN signal at mid cell was 0.63, 0.56 and 0.44 respectively. Using Eq. 1 and [(Td − tx)/Td] × 100 it was calculated that the FtsZ-ring appears after 29% and that PBP3 and GFP-FtsN localize at mid cell after 36% and 47% of the cell cycle respectively (Table 2 and Fig. 3). Constrictions were visible after 55% of the cell cycle (Table 2 and Fig. 5). The divisome maturation time was the same in the LMC500 as in the LMC500pDML2499 (expressing GFP-FtsN) culture (20% and 18% of the cell cycle respectively). GFP-FtsN overproduction does not seem to be able to speed up divisome maturation. However, the expression of GFP-FtsN caused a delay in the synthesis of the new cell poles because in the wild-type strain, constriction required 21% of the cell cycle whereas in the presence of GFP-FtsN division required 45% of the generation time. To verify that the GFP-part of the fusion protein was not responsible for the prolonged constriction time, LMC500 was transformed with plasmids pDML2001 or 2000 (Table 1). These plasmids expressed FtsN constitutively at a fourfold and 25-fold concentration (Fig. 6), respectively, compared with endogenous FtsN (see Experimental procedures on the quantification of FtsN). The overexpression of FtsN without the GFP part also caused a increase in the duration of the constriction period (Table 2). Therefore, GFP is not the cause of the prolonged constriction period.
Because the delay in DNA segregation in LMC510 seemed to be correlated with a prolonged constriction period, the prolonged constriction period of FtsN overexpressing cells might be correlated with a delay in DNA segregation. Therefore, the nucleoid segregation time in LMC500, LMC500 pDML2499 expressing GFP-FtsN, LMC500 pMD2001 over expressing FtsN fourfold, and LMC500 pDML2000 over expressing FtsN 25-fold was determined from fluorescence images of DNA (DAPI) stained cells. Segregated nucleoids could be observed during the last 35% and 39%, 41%, and 27% of the cell cycle of the last four strains mentioned respectively. Like in LMC510, the prolonged constriction period in strains overexpressing FtsN seemed to coincide with a delay in nucleoid segregation and not with a delay in divisome maturation. These results suggest again a correlation between new cell pole synthesis and DNA segregation.
Time delay between FtsZ-ring assembly and divisome maturation
In a previous study (Den Blaauwen et al., 1999) the time between the initiation of cell division by the polymerization of FtsZ in a ring at mid cell and the appearance of a constriction visualized by electron microscopy was determined to last approximately 20 min in E. coli cells growing with a Td time of 85 min This invokes the question what is happening in between the formation of the FtsZ-ring and the actual constriction process. To determine whether divisomal proteins assemble simultaneously or sequentially in time, we studied the time between the assembly of the FtsZ-ring and the arrival of cell division protein FtsQ, FtsW, PBP3 and FtsN.
For a Td time of 85 min, a model emerges in which cell division occurs in two steps. The first step is the FtsZ-ring polymerization at one or two sites of the cytoplasmic membrane at mid cell, more or less simultaneously with termination of DNA replication. While the ring grows ZipA, FtsA and ZapA assemble onto it. Whether FtsE/X and FtsK localize during this process or should be grouped with FtsQ and its downstream protein is not yet clear. In the second step after approximately 17 min FtsQ, FtsW, PBP3 and FtsN assemble in perhaps 1–3 min (Fig. 7). Although, we have only determined the timing of four division proteins in relation to the timing of FtsZ localization, their interdependence for localization suggests that all proteins downstream of FtsK assemble simultaneously (Fig. 7). The timing of AmiC localization was not determined. However, based on its dependency on the localization of FtsN, it seems unlikely that the protein would localize earlier than FtsN.
Recently, Morlot et al. (Morlot et al., 2003) analysed the localization of a number of cell division proteins in Streptococcus pneumonia by immunofluorescence microscopy. They showed that FtsZ localized at the equatorial plane of the future daughter cell, well before PBP1a (PBP1a/1b in E. coli), PBPB2x (PBP3 in E. coli) and FtsW became localized at the same position. This suggests the presence of a time delay between Z-ring formation and the arrival of other cell division proteins in this gram-positive organism as well. They also showed that in the absence of carboxypeptidase PBP3 (PBP5 in E. coli), the FtsZ-ring could be completely dissociated from a putative ring of high molecular weight PBPs. The results of Morlot et al. (Morlot et al., 2003; 2004) show that the FtsZ-ring and the peptidoglycan synthesizing enzymes can form separate entities. Using a plasmid-encoded two-hybrid system in which a large number of cell division proteins were systematically paired, FtsZ was reported to interact with FtsK and FtsA was reported to bind to PBP3 (Di Lallo et al., 2003). All other interactions occurred within a group containing FtsZ, FtsA, ZipA and ZapA or within a group containing FtsK and downstream proteins (FtsE/X was not assayed). Because these interactions were identified under conditions of at least several-fold overproduction, not all interactions might exist under physiological conditions. The observation of Di Lallo et al. (2003) that the interactions can be grouped in a inhomogeneous FtsZ, FtsA, ZipA cluster I and the other proteins in a homogenous cluster II, provides also an indication for the possible existence of two distinguishable protein complexes.
The average number of FtsQ proteins per cell is in the order of 20–40 (Carson et al., 1991) and for instance for PBP3 the number is approximately 100 (Dougherty et al., 1996). Therefore, a limited number of protein complexes involved in envelope synthesis can be expected instead of a ring-like structure composed of these proteins. We found that an average cell contains about 1000 mol FtsN, which is enough to fill the circumference of the cytoplasmic membrane in the periplasm. FtsN might assist in the positioning of the cell division proteins that are present in fewer numbers. Recently, Ursinus et al. (2004) showed that the carboxy-terminal domain of FtsN is able to bind to glycan strands that are preferably about 25 dissacharide units in length. Although the specific FtsN domain appeared to be dispensable for cell division, it might add to the stability of the positioning of the divisome.
Does the divisome maturation time delay have a function?
Why is there a time delay between Z-ring assembly and the arrival of FtsQ up to FtsN? At least two cell cycle-related activities can be discriminated during the time delay. Because the Z-ring assembly coincides with the termination of DNA replication, the ongoing DNA segregation is completed at some point between Z-ring formation and cell division. We compared the time delay of cell division proteins that had a normal DNA segregation period and a delayed DNA segregation period. We did not find a correlation between the duration of the time delay and the DNA segregation period. Therefore, it appears unlikely that the time delay is required for the segregation of the nucleoids.
The second cell cycle related activity is the synthesis of polar peptidoglycan. Below we provide three arguments that favour a role for FtsZ in the switch from cylindrical to polar peptidoglycan synthesis.
First, Wientjes and Nanninga (1989) determined the topography of [3H]-Dap incorporation by pulse labelling and autoradiography. They deduced a switch from lateral to central synthesis and the increased synthesis in the cell centre was at the expense of synthesis in the lateral wall. The increase of central synthesis started approximately after 42% of the cell cycle, which is at about the same time as the appearance of the Z-ring (Woldringh et al., 1987; Den Blaauwen et al., 1999). The increase reached a maximum in constricting cells and it did not change as the constriction progressed to a more advanced state. The increase in central peptidoglycan synthesis was not dependent on the localization or the function of PBP3 (Wientjes and Nanninga, 1989). Autoradiography of FtsZ filaments that had incorporated [3H]-Dap showed a random distribution of silver grains with no sign of an increased incorporation at a specific place on the envelope (Woldringh et al., 1987). Thus, positioning of the FtsZ-ring seemed to attract central peptidoglycan synthesis.
Second, De Pedro et al. (De Pedro et al., 1997) replaced the diaminopimelic acid [3H]-Dap by d-cystein. Subsequently, the d-cysteines were biotinylated in isolated sacculi and their position detected by antibiotin antibodies conjugated to gold particles. This resulted in a dense labelling of the sacculi. By chasing the label for one or two Tds, the incorporation of new peptidoglycan could be followed by the dilution of gold particles. A main conclusion was that the new cell poles consisted of new peptidoglycan, whereas old cell poles were completely inert with respect to peptidoglycan synthesis. In line with the experiments of Wientjes and Nanninga (1989), they also showed that PBP3 filaments contain bands of newly synthesized peptidoglycan of approximately the surface of two new cell poles, indicating that PBP3 was not involved in the insertion of new material. However, FtsZ filaments did not contain these bands but consisted completely of newly inserted peptidoglycan in old peptidoglycan, which is the cylindrical type of peptidoglycan synthesis. Apparently, the presence of FtsZ is required for the synthesis of polar peptidoglycan. In a recent article Varma and Young (Varma and Young, 2004) also provided some evidence for the idea that FtsZ is somehow involved in peptidoglycan metabolism.
Third, the FtsZ mutant (FtsZR174D) is defect in its association with the membrane, but is able to localize in vivo (Koppelman et al., 2004) It is able to polymerize but can not bundle in vitro. Remarkably, FtsZR174D filaments have blunt constrictions instead of the usual smooth FtsZ filaments in which the Z-ring is not formed. This mutant is not able to bind any of the proteins that localize downstream of FtsK (D. Blaauwen unpubl. results).
All the above mentioned observations suggest that FtsZ is able to initiate the process of cell division by facilitating somehow the switch from lateral peptidoglycan synthesis to the increased central synthesis that consists of completely new peptidoglycan. It should be noted that the period in which the Z-ring is present is approximately equal to the polar surface to be synthesized at the two slow growth rates analysed in this article.
What might be the role of PBP3?
PBP3 and FtsQ filaments also show blunt constrictions (Taschner et al., 1988). PBP3 nor FtsQ seems to be required for this initial peptidoglycan synthesis. This could indicate that all proteins that assemble approximately 17 min after the localization of the Z-ring under our growth conditions might not be involved in this peptidoglycan synthesis switch from lateral to central growth. PBP3 does not have transglycosylase activity but only transpeptidase activity (Adam et al., 1991; Nguyen-Distèche et al., 1998) and the transglycosylase activity seems to be a prerequisite for the insertion of peptidoglycan precursors in the existing layer (Den Blaauwen et al., 1990). Therefore it seems plausible that PBP3 is not inserting peptidoglycan precursors, but is modifying the already synthesized polar peptidoglycan that has been synthesized by other synthetases during the time delay. It is feasible that the proteins that localize after the time delay are all involved in modification of the cylindrical cell envelope in to a polar envelope, without actually synthesizing new peptidoglycan. Inhibition of PBP2 or PBP1a/b does not prevent constriction (Wientjes and Nanninga, 1989). For these reasons it has been argued that a Penicillin Insensitive Peptidoglycan Synthesis (PIPS) step must precede constriction of the envelope (Nanninga, 1991). A possible candidate for the synthesis of the new polar peptidoglycan could be the monofunctional transglycosylase encoded by the mgt gene (Di Berardino et al., 1996) because it is able to use the lipid-II precursor as substrate. This transglycosylase activity would result in local uncross-linked or weakly cross-linked peptidoglycan, depending on whether one of the Class A PBPs joins the peptidoglycan synthesis. This weakly cross-linked material could then easily be modified in shape by the activity of PBP3/PBP1B.
In conclusion, our results provide evidence that the Z-ring is somehow involved in the initiation of synthesis of polar peptidoglycan. After 17 min of further growth cell division proteins downstream of FtsK assemble. Subsequently, the latter proteins might determine the shape of the cell pole (Fig. 7). The driving force for the assembly of these proteins might be the presence of suitable substrate i.e. weakly cross-linked peptidoglycan.
Bacterial strains and growth conditions
Escherichia coli K12 cells were grown steady state at 28°C in glucose minimal (GB1) medium containing 6.33 g of K2HPO4·3H2O, 2.95 g of KH2PO4, 1.05 g of (NH4)2, 0.10 g of MgSO4·7H2O, 0.28 mg of FeSO4·7H2O, 7.1 mg of Ca(NO3)2·4H2O, 4 mg of thiamine, 4 g of glucose and 50 µg of required amino acids per litre pH 7.0. LMC500, LMC509, LMC510 and LMC531 require lysine, for growth in minimal medium (Table 1). LMC509 was grown in 1/2 GB1 (Koppelman et al., 2004).
All E. coli strains were grown at 28, 37, or 42°C in rich medium containing 10 g of bactotryptone, 5 g of yeast extract, 5 g of NaCl, 15 mmol NaOH per litre (TY) or 10 g of NaCl (LB). When required (Table 1), 100 µg ml−1 ampicillin or 50 µg ml−1 chloramphenicol was added to medium. Absorbance was measured at 450 nm (GB1), or at 600 nm (LB and TY) with a 300-T-1 spectrophotometer (Gilford Instrument Laboratories). Cell numbers were monitored using an electronic particle counter (orifice 30 µm). Cultures were considered to be in steady state of growth if the ratio between optical density and number of cells remained constant over time (Fishov et al., 1995).
Determination of the average cell age at which the divisome
The fraction F(x) of cells with a particular characteristic, that occurs at the end of the cell cycle (e.g. constricting, Z-ring containing or FtsN localizing) contains all cell ages at which the cells have this characteristic. By determining the fraction of cells with the before mentioned characteristics, the age at which these characteristics appears in the cell cycle can be calculated using:
where Td is the mass doubling time and tx the time (min) during which cells have a certain characteristic at the end of the cell cycle and which lasts until the separation into two daughter cells (Powell, 1956).
Isolation of FtsN and production of polyclonal antiserum against FtsN
HB101 strain was transformed by the pKD160 (Dai et al., 1996) carrying the ftsN gene coding for a soluble form of FtsN. Cells were cultured at 37°C in LB medium containing 100 µg ml−1 ampicillin until an absorbance (600 nm) of 0.6 was reached. The culture was supplemented with 1 mM isopropyl β- d-thiogalactopyranoside (IPTG) for 3 supplementary hours. The cells were spheroplasted and the periplasmic fraction collected (Fraipont et al., 1994) and dialysed against 20 mM sodium phosphate buffer pH 7.0. The protein was purified on a SP sepharose HP XK 26/10 cation exchange column (Amersham). The protein (1.5 mg, 90% pure) was laid on a 10% sodium dodecyl sulphate polyacrylamide gel (SDS-PAGE) and run for 20 h at 4°C. A part of the gel was blotted on an Immobilon Millipore PVDF membrane to verify the N-terminal sequence of the 30 kDa protein (Van Beeumen Gent). The corresponding FtsN slice was cut off, ground in 20 mM sodium phosphate, 50 mM NaCl, 0.1% SDS in a dry ice refrigerated mortar, centrifuged for 20 min at 5000 g and injected in rabbits (Gamma Belgium). The antibodies (dilution 10 000-fold) did not show any unspecific reaction on an immunoblot containing a total extract of E. coli (12 µl of a culture at an OD600 of 1.2) after SDS-PAGE.
Quantification of FtsN
LMC500 and LMC500 transformed by the FtsN and GFP-FtsN expressing plasmids pDML2000, pDML2001, pDML2499 and pDML2002 were grown to steady state in GB1 at 28°C. At an OD450 of 0.2 the cells were harvested and applied on an SDS-PAGE. Optical density was measured using a 300-T-1 spectrophotometer (Gilson Instruments Laboratories). Cell numbers were monitored using an electronic particle counter (orifice 30 µm). The amount of FtsN was estimated by submitting samples of the purified protein to SDS-PAGE and by performing density measurements of the Coomassie-stained 30 000 Da protein band. BSA was used as reference. The protein concentration of each sample was chosen such that similar amounts of FtsN per sample were applied to ensure equal efficiency in blotting and antibody binding. The blots were developed with a polyclonal antibody against FtsN and a chemiluminescence kit (Tropix, Bedfor, MA). The number of endogenous FtsN protein per cell was determined using the purified soluble form of FtsN as internal standard for the quantification by densitometry. A series of FtsN concentrations of 0.8–1.4 ng FtsN was applied in duplo on an SDS-PAGE that also contained in duplo lanes with 1 µl, 2.5 µl and 5 µl of LMC500 cells harvested at an optical density of 0.2 at 450 nm. An immunoblot obtained from the gel was developed by the polyclonal antiserum against FtsN and chemiluminescence. The densitrometric result of each duplo of the soluble FtsN samples were averaged and plotted against the amount of FtsN (Fig. S1). Linear regression resulted in a calibration line y = (0.13 ± 0.03) x = (0.08 ± 0.03). Subsequently the average density of the FtsN band in the LMC500 bands was determined and the number of FtsN molecules per microlitre of LMC500 cells was calculated to be 0.756 ± 0.122 ng using the calibration line. Under the described conditions, which are identical to the condition in the timing of the localization of cell division proteins experiments, the number of FtsN molecules per average cell in the culture was 4000 for pDML2499, pDML2002 (both producing GFP-FtsN) and pDML2001 (producing FtsN) and 25 000 for pDML2000 (producing FtsN) respectively [number of experiments (n) = 8]. The number of endogenously produced FtsN was determined to be 1000 mol per average cell (n = 6). The amount of FtsQ produced by LMC500, LMC2328 and LMC500pTHV039 was determined using the same experimental approach. Because we did not have purified FtsQ, the amount of FtsQ expressed was related to the amount of endogenously expressed FtsQ. The LMC2328 produced GFP-FtsQ and the plasmid pTHV039 encoded GFP-FtsQ were produced in five- and 70-fold excess (n = 4), respectively, compared with the endogenous amount of FtsQ produced.
Expression of GFP-fusion proteins and plasmid construction
For the localization studies, E. coli FtsN, protein was expressed as an N-terminal GFP fusion protein. A three Asn linker separated the GFP and the amino terminus of the FtsN protein. The ftsN gene was PCR amplified using Pfu DNA polymerase (Stratagene, La Jolla), genomic DNA from E. coli LMC500 and the sense primer 5′-TAGAATTCAACAACAA CGTGGCACAACGAGATTATGTACGC-3′ and antisense 5′-TAT AAGCTTGCAATT A T AGATGGGGGGGATTTTGAGG-3′. The PCR fragment was then digested with EcoRI/HindIII (underlined sites) and ligated into cleaved pTHV038 vector to form pDML2002. To modify the replication origin of pDML2002, a fragment containing the P15a replication origin and chloramphenicol resistance gene was amplified by PCR using pMCL210 as template and the sense primer 5′-TATTG CATGCCTGGGGTGCCTAATGAGTGAGC-3′ and antisense primer 5′-TAAAAAGCTTAAATCCTGGTGTCCCTGTTGATA CCG-3′. This fragment was digested by HindIII/SphI and exchanged with the SphI/HindIII (underlined sites) fragment from pDML2002. The resulting plasmid was called pDML2499. The ftsN sequencing revealed the same modification as described above. These vectors contain the weakened trc promoter as described by Chen et al. (1999) in which GFPmut2 (Cormack et al., 1996) has been cloned (Table 1). Cells of various strains were transformed and produced GFP-FtsN without induction with IPTG if not specifically mentioned.
Plasmid pDML2001 expresses FtsN under control of the trc promoter weakened in the −35 region. The encoded FtsN has an additional peptide MEFNNN at its amino end. The EcoRI-HindIII fragment carrying ftsN was excised from the pDML 2499 and exchanged with the EcoRI-HindIII fragment carrying the gfp gene from pDSW208 (Weiss et al., 1999) to create pDML2001.
Plasmid pDML2000 expresses FtsN was under the control of the lac promoter. The ftsN gene was amplified by PCR using E. coli Top 10 as template, 5′-CGGAATTCGGATCC TGAT ACAGCGAAACGA T AGTGGCACAACGAGA TTATGTA CGC-3′ as sense primer and 5′-GCATGCGATCGGTACGC CAGTCTTTTGCC-3′ as antisense primer. The sequence upstream from the start codon (in bold) carries the ribosome binding site of ftsN. The purified PCR fragment was digested by BamHI and PvuI (underlined sites) and ligated into the same sites of pMCL210 containing the p15A origin. The BamHI-PvuI DNA segment was completely sequenced. In the resulting plasmid called pDML2000.
The ftsW gene was amplified by PCR using pDML2400 as template, 5′-CACGAATTCAACAACAACCGTTTATCTCTCCC TCGCCTGAAAATGC-3′ as sense primer and 5′-GCCGC AAGCTTATCATCGTGAACCTC-3′ as antisense primer. The purified PCR fragment was digested by EcoRI and HindIII (underlined sites) and ligated into the same sites of pDSW234. The EcoRI-NruI DNA segment was completely sequenced. The NruI-HindIII DNA fragment was then replaced by the NruI-HindIII DNA fragment from the original pDML2400 to create pDML2414. pDML2400 is a pET28a(+) derivative containing the wild-type ftsW gene. The gene fusion was under the control of the trc promoter weakened in the −35 region and encoded FtsW fused to the carboxy-terminus of the GFP.
Cells were fixed and permeabilized as described (Den Blaauwen et al., 2001). Incubation with the polyclonal against FtsN, affinity purified polyclonal antibodies PBP3 (Marec-Fairley et al., 2000) or Fab fragments of monoclonal antibody F168-12 against FtsZ (Voskuil et al., 1994) conjugated with AlexaTM 546 or AlexaTM 488 (Molecular Probes), diluted in blocking buffer was carried out for 60 min at 37°C as described. When IgG was used, incubation with secondary antibodies, goat anti-rabbit conjugated with AlexaTM 546, diluted in blocking buffer was carried out for 30 min at 37°C (Den Blaauwen et al., 2001).
We thank Nanne Nanninga for critically reading the manuscript. We thank David S. Weiss for the gift of plasmid pDSW208 an plasmid pDSW234. This work was supported in part by the Belgian program on Interuniversity Poles of Attraction initiated by the Belgian State, Prime Minister's Office, Services fédéraux des affaires scientifiques, techniques et culturelles (PAI no. P5/33), the Fonds de la Recherche Fondamentale Collective (Contract No. 2.4521.01) and a Vernieuwingsimpuls Grant 016.001.024 (T. den B.) Netherlands Organization for Scientific Research (NWO). This work was partially funded by the sixth European Framework Program (COBRA LSHM-CT-2003-503335). A. P. is a fellow of the Fonds pour la Formation à la Recherche dans l’Industrie et l’Agriculture, Brussels.