Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
In most bacteria cell division is mediated by a protein super-complex called the divisome that co-ordinates the constriction and scission of the cell envelope. FtsZ is the first of the divisome proteins to accumulate at the division site and is widely thought to function as a force generator that constricts the cell envelope. In this study we have used a combination of confocal fluorescence microscopy and fluorescence recovery after photobleaching (FRAP) to determine if divisome proteins are present at the septum at the time of cytoplasmic compartmentalization in Escherichia coli. Our data suggest that many are, but that FtsZ and ZapA disassemble before the cytoplasm is sealed by constriction of the inner membrane. This observation implies that FtsZ cannot be a force generator during the final stage(s) of envelope constriction in E. coli.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Cell division in the Gram-negative bacterium Escherichia coli requires constriction and eventually scission of the cell envelope. This process is mediated by a protein super-complex (the ‘divisome’), which consists of more than 24 different proteins (reviewed in Vicente et al., 2006; de Boer, 2010; Lutkenhaus et al., 2012). The assembly of these proteins is hierarchical and occurs in a two-step process (Aarsman et al., 2005), starting with the arrival of FtsZ at the mid-cell (reviewed in Adams and Errington, 2009; Erickson et al., 2010). FtsA and ZipA are simultaneously recruited, and tether FtsZ to the inner membrane (Pichoff and Lutkenhaus, 2002). Four non-essential divisome proteins (i.e. ZapA, ZapB, ZapC and ZapD) are also recruited by FtsZ-FtsA-ZipA, to form an intermediate structure called the Z-ring (Adams and Errington, 2009; Durand-Heredia et al., 2011; 2012; Hale et al., 2011). In the second phase of assembly, inner membrane, periplasmic and outer membrane proteins are recruited to form the mature divisome (Aarsman et al., 2005; Vicente et al., 2006; Gerding et al., 2007; de Boer, 2010; Typas et al., 2010; Lutkenhaus et al., 2012). Once assembled, the divisome constricts the cell envelope and ultimately brings about the sequential closure of the cytoplasm and then the periplasm (Skoog et al., 2012).
FtsZ is a tubulin-like GTPase that is thought to play two key roles in the division process. First as an assembly platform for other cell division proteins, and then as a force generator that pulls the cell envelope inward (Adams and Errington, 2009; Erickson et al., 2010; Mingorance et al., 2010). The latter role is more controversial and is founded on two key observations: (i) FtsZ-YFP can constrict liposomes when fused to a membrane targeting sequence (Osawa et al., 2008; 2009), and in the presence of FtsA* these constrictions occasionally go to completion (Osawa and Erickson, 2013); and (ii) FtsZ filaments can undergo a conformational change from straight to curved in vitro (Lu et al., 2000). Curiously however, calculations derived from the in vitro analysis suggest that FtsZ proto-filaments can only curve to a diameter of ∼ 24 nm. If the tethers provided by FtsA and ZipA are taken into account, then this calculation implies that FtsZ proto-filaments can only pull the inner membrane to a diameter of ∼ 57 nm (not to complete closure) (Erickson et al., 2010). In this study we provide evidence that FtsZ departs from the septum before the cytoplasmic compartment has been separated. This simple observation implies that FtsZ cannot constrict the inner membrane during the final stage(s) of septal closure.
FtsZ–GFP disassembles from the divisome prior to closure of the cytoplasm
We engineered a strain of E. coli (MG1655) that simultaneously expressed three different fluorescent proteins; Cerulean in the cytoplasm (CeruleanCYTO), mCherry in the periplasm (mCherryPERI) and FtsZ–GFP. The ratio of native FtsZ: FtsZ–GFP was approximately 60:40 and the engineered strain grew similarly to the parental strain (Fig. S1), indicating that cell division was not perturbed. Visual inspection of the cells by Super-Resolution Structured Illumination Microscopy (SR-SIM; Gustafsson, 2000) (Fig. 1A) and confocal microscopy (Fig. 1B) confirmed that there were different stages of FtsZ–GFP localization during the cell cycle, as reported by others (Sun and Margolin, 1998; Den Blaauwen et al., 1999; Wang et al., 2005). Briefly, FtsZ–GFP was first identified at the division septum as two dots or a bar (stage 1). The two dots then converged to a single dot (stage 2). In a third, more transient stage FtsZ–GFP was no longer visible at the division septum (stage 3). Time-lapse imaging indicated that this stage represented roughly 1–2% of the cell cycle (Fig. S2). Finally, in a fourth stage, FtsZ–GFP re-localized to the future division site of the daughter cells (stage 4). To correlate the four stages of FtsZ–GFP localization with envelope constriction we combined the confocal microscopy with fluorescence recovery after photobleaching (FRAP) (Stromqvist et al., 2010; Skoog et al., 2012). In the FRAP experiment we irreversibly bleached both CeruleanCYTO and mCherryPERI on one side of the division septum and monitored the diffusion of fluorescence from the non-bleached half of the cell over time (Fig. 1B). Fluorescence recovery therefore indicated whether the cytoplasm and/or periplasm were still connected across the division septum (open), or whether they had been compartmentalized by the division septum (closed). In 186/188 cells where FtsZ–GFP was visualized at the division septum (stages 1 and 2), both CeruleanCYTO and mCherryPERI were able to diffuse to the bleached half of the cell thus indicating that both compartments were open (Fig. 1C). In 22/33 cells where FtsZ–GFP was absent from the division septum (stage 3), the FRAP measurement indicated that the cytoplasm and the periplasm were still open (Fig. 1C). However, in the remaining 11/33 cells the cytoplasm was closed across the division septum, and the periplasm was either open or closed (5 and 6 cells respectively). All cells in this stage were followed from a stage where FtsZ–GFP was accumulated at the septum to a stage where it was not, to ensure sequential progression of the FtsZ–GFP localization pattern. Together, these data indicate that the cytoplasmic and periplasmic compartments are closing (or compartmentalizing) at the stage when FtsZ–GFP is no longer detected at the divisome. By the time FtsZ–GFP had relocated to the future division sites in the daughter cells (stage 4), the cytoplasm and the periplasm were both closed across the division septum (Fig. 1C).
Most divisome proteins remain at the division site until after the cytoplasm is closed
To determine if other divisome proteins stayed longer at the division site we expressed GFP–ZapA, FtsA–GFP, ZipA–GFP, GFP–FtsQ, GFP–FtsL or GFP–FtsI together with mCherryCYTO (FtsZ–GFP was included as a control). The ZipA–GFP, GFP–FtsQ, and GFP–FtsI fusions were expressed at lower levels than their native counterparts (Fig. S1) and all strains grew similarly to the parental strain indicating that division was not perturbed (Fig. S1). The arrival and departure of all GFP fusions at the division septum was similar to FtsZ–GFP (Fig. 2A) and could be correlated with envelope constriction using the FRAP approach (Fig. 2B). These data indicated that the transition from ‘cytoplasm-open’ to ‘cytoplasm-closed’ occurred while FtsA–GFP, ZipA–GFP, GFP–FtsQ, GFP–FtsL and GFP–FtsI were present at the division septum (stage 2) (n > 150 for each strain) (Fig. 2B). In contrast GFP–ZapA was absent from the division septum during this transition (stage 3).
FtsZ-mCherry disassembles from the divisome before other divisome proteins
The FRAP data indicated that FtsZ–GFP and GFP–ZapA disassembled from the division site prior to compartmentalization of the cytoplasm, whereas other divisome proteins remained until the cytoplasm compartmentalized. To substantiate this observation we monitored the localization of FtsZ-mCherry along with GFP–ZapA, ZipA–GFP, FtsA–GFP, GFP–FtsL, GFP–FtsQ or GFP–FtsI (n > 150 for all strains; Fig. 3). Again, all strains grew similarly to the parental strain, indicating that division was not perturbed (Fig. S3). Notably, FtsZ-mCherry and GFP–ZapA colocalized throughout the cell cycle (n = 792). This observation was expected, as ZapA is associated with the inner part of the Z-ring and should thus serve as a marker for FtsZ (Galli and Gerdes, 2010). For all other strains the mCherry and GFP signals also colocalized at the septum through most of the constriction process, however, at a late stage the mCherry signal disappeared from the division septum while the GFP signal remained. Those GFP fusion proteins that remained at the division septum were either membrane anchored (FtsA) or embedded (ZipA, FtsQ/L/I). In cells expressing GFP–FtsL, GFP–FtsQ or GFP–FtsI (but not ZipA–GFP or FtsA–GFP), the GFP signal remained at the division septum even when FtsZ-mCherry had relocated to the daughter cells. As expected all GFP fusions eventually relocated to the daughter cells to again colocalize with FtsZ-mCherry. Although we did not monitor FtsK in our experiment, Sherratt and co-workers have noted that FtsK-YFP remains at the division site longer than FtsZ-CFP (Wang et al., 2005). Taken together with the FRAP data, we conclude that FtsZ–GFP and GFP–ZapA disassemble from the division site before other division proteins. This event occurs before compartmentalization of the cytoplasm.
The localization patterns of FtsZ–GFP (and FtsZ-mCherry) reflect those of FtsZ
Although fluorescent fusions to FtsZ are not fully functional in vivo (Ma et al., 1996), they can complement an ftsZ84(Ts) mutant (Weiss et al., 1999) and their localization patterns and dynamics mimic that of the native FtsZ (see (Margolin, 2012) and references therein). To further substantiate the latter point we monitored the localization patterns of both FtsZ and GFP in a strain expressing both the native FtsZ and the FtsZ–GFP fusion by immuno-cytochemical fluorescence microscopy. In all cells analysed (n = 178) FtsZ colocalized with GFP, indicating that the native FtsZ behaved in a similar manner to FtsZ–GFP (Fig. S4). One notable limitation with IFM is that it is not possible to discriminate between cells that are in stage 3 (i.e. the protein is no longer at the septum) from cells that have not labelled with the antibodies.
FtsZ is thought to be a major force generator that pulls the inner membrane towards closure during division in E. coli (Adams and Errington, 2009; Erickson et al., 2010; Mingorance et al., 2010). To better understand this role, we correlated the localization of FtsZ–GFP at the division septum with envelope constriction by dual colour FRAP. Intriguingly, we noted that FtsZ–GFP disassembled from the division septum before the cytoplasmic and periplasmic compartments were sealed. To determine if other cell division proteins behaved in a similar manner to FtsZ–GFP we monitored the localization of GFP–ZapA, ZipA–GFP, FtsA–GFP, GFP–FtsL, GFP–FtsQ and GFP–FtsI during cytoplasmic compartmentalization. GFP–ZapA behaved in a similar manner to FtsZ–GFP, departing from the divisome before cytoplasmic compartmentalization, while all other proteins remained until the cytoplasm had been compartmentalized. The stepwise disassembly of the divisome was verified by carrying out dual colour fluorescence imaging. Although we have not assayed all divisome proteins in this study, other groups have noted that fluorescently labelled FtsZ, ZapA and ZapB colocalized throughout the cell cycle, but that FtsK remained longer at the division septum (Wang et al., 2005; Galli and Gerdes, 2010). Collectively, these observations are consistent with the notion that late stages of inner membrane constriction do not involve FtsZ or ZapA.
One caveat to our interpretation is that we do not know the detection limit for FtsZ–GFP. This limit will change during constriction as the amount of divisome bound FtsZ–GFP transits from approximately 33% to 0%. The possibility therefore remains that FtsZ–GFP is present in the deepest constrictions but not detected. However, this scenario seems unlikely as many other GFP–Fts fusion proteins were readily detected in deep constrictions even after FtsZ–GFP had disappeared. Some of these proteins, like GFP–FtsI and GFP–FtsQ, were more challenging to visualize than FtsZ–GFP on account of their significantly lower abundance, as documented by Western blotting and comparatively weak fluorescence during microscopy. Additionally, FtsZ–GFP can be detected at the new pole after division in Caulobacter crescentus and Agrobacterium tumefaciens (Thanbichler and Shapiro, 2006; Zupan et al., 2013), implying that it is possible to detect it in deeply constricted cells if it is still present.
Another consideration in the interpretation of our data is whether a GFP fusion protein reflects the behaviour of the native protein in every detail. GFP–ZapA, GFP–FtsQ, GFP–FtsL and GFP–FtsI fusions can complement their respective null mutants indicating that they are fully functional (Chen et al., 1999; Ghigo et al., 1999; Weiss et al., 1999; Galli and Gerdes, 2010). It therefore seems reasonable to assume that the localization patterns of these fusions are similar to the native protein. FtsZ–GFP, FtsA–GFP, and ZipA–GFP cannot fully complement their respective null mutants but their fluorescence localizations patterns and dynamics are thought to follow those of the native versions (Ma et al., 1996; Hale and de Boer, 1997; Margolin, 2012 and references therein). For FtsZ–GFP we re-evaluated this point by immuno-cytochemical fluorescence microscopy. Considering all the data, we suggest that the native FtsZ and ZapA also dissociate from the division site before cytoplasmic compartmentalization, while native FtsA, ZipA, FtsQ, FtsL and FtsI do not (Fig. 4). The disassembly of FtsZ–GFP before cytoplasmic compartmentalization was unexpected, as Osawa and Erickson have noted that FtsZ-YFP and FtsA* can (in rare cases) divide liposomes (Osawa and Erickson, 2013). However, it is worth noting that there are no negative-regulatory systems (like Noc and MinCDE) acting on FtsZ-YFP in the liposomes, so it is possible that FtsZ proto-filaments persist longer than they would do in vivo.
Our data do not enable us to accurately determine when FtsZ–GFP and GFP–ZapA disassembled, but we estimate that it occurs at a diameter > 12 nm. This estimation is based on calculations that GFP can freely diffuse through a ring-diameter of > 12 nm (Stromqvist et al., 2010). Our estimate is consistent with in vitro experiments and calculations done by Erickson and colleagues (Erickson et al., 1996; 2010; Lu et al., 2000; Chen and Erickson, 2009). They estimate that highly curved FtsZ proto-filaments could bend to a minimum diameter of ∼ 24 nm, which would correspond to a diameter of ∼ 57 nm for the inner membrane (Erickson et al., 2010). In this scenario FtsZ would be unable to complete the closure of the inner membrane. We speculate that disassembly of FtsZ (and ZapA) from the leading edge of the constricting inner membrane might occur so that fusion of the lipid bilayer can proceed. In principle, fusion could be driven by lipid synthesis, as shown in the L-form of Bacillus subtilus (Mercier et al., 2013), or by inward growth of the peptidoglycan. Further work is required to gain insight into this fundamental aspect of division.
Strain and plasmid construction
The locus encoding the GFP fusions from strains EC448 (MC4100 Δ[λattL-lom]::bla lacIqP208-ftsZ–gfp), EC450 (MC4100 Δ[λattL-lom]::bla lacIqP208-zipA–gfp), EC447 (MC4100 Δ[λattL-lom]::bla lacIqP210-ftsA–gfp), EC442 (MC4100 Δ[λattL-lom]::bla lacIqP207–gfp–ftsQ), EC438 (MC4100 Δ[λattL-lom]::bla lacIqP207–gfp–ftsL) and EC436 (MC4100 Δ[λattL-lom]::bla lacIq P207–gfp–ftsI) (Chen et al., 1999; Ghigo et al., 1999; Weiss et al., 1999) were transferred to MG1655 by P1 phage transduction, selecting for AmpR, to create strains BS001, BS002, BS003, BS004, BS005 and BS007 respectively (Table S1).
The gfp–zapA fusion was constructed by PCR amplifying zapA from E. coli MG1655 chromosomal DNA using primers zapA_up (CACGAATTCAACAACAACATGTCTGCACAACCCGTCGATATC) and zapA_down (CACAAGCTTCATTCAAAGTTTTGGTTAGTTTTTTCG). The 356 bp product was cut with EcoRI and HindIII (sites underlined in primers) and ligated into the GFP fusion vector pDSW207 (Weiss et al., 1999). The resulting plasmid was verified by sequencing and designated pDSW1646. Next, pDSW1646 was digested with SphI and ScaI to obtain a ∼ 3.4 kb restriction fragment carrying lacIq and P204::gfp–zapA. This fragment was ligated into SphI/ScaI-digested pJC69 (Chen and Beckwith, 2001) to produce pDSW1655. Finally, plasmid pDSW1655 was integrated into the phi 80 phage attachment site of MG1655 as described (Haldimann and Wanner, 2001). The final strain was designated EC3408 and has the following genotype: MG1655 Φ80::pDSW1655(lacIq P204::gfp–zapA SpcR).
The pRha-BAD plasmid was constructed by PCR amplifying the multiple cloning site of pBAD24, and cloning it into a non-coding region of the pRha67 plasmid (between the ColE1 origin of replication and the multiple cloning site). The pRha67 (Giacalone et al., 2006) plasmid was obtained from Jan Willem de Gier (Stockholm University). The pKS1 plasmid was constructed by cloning the gene for the Cerulean fluorescent protein (Rizzo et al., 2004) downstream of the rhamnose inducible promoter in pRha-BAD. The gene encoding mCherry (Shaner et al., 2004) was fused to the region encoding the signal sequence of TorA as described previously (Skoog et al., 2012), and cloned downstream of the arabinose inducible promoter in pRha-BAD. The pKS2 plasmid was constructed by cloning the gene for mCherry downstream of the rhamnose inducible promoter in the pRha67 plasmid. All cloning was carried out using the uracil-excision method (Norholm, 2010). Both constructs were verified by DNA sequencing (MWG, Germany). A complete list of the strains and plasmids used in this study is provided in Table S1.
Bacterial growth and fluorescent protein expression
To create the strain expressing FtsZ–GFP, CeruleanCYTO and mCherryPERI, we transformed BS001 with pKS1. Cells were grown as described previously (Thomas et al., 2001; Skoog et al., 2012), with the addition of 2.5 μM Isopropyl β-D-1-thiogalactopyranoside (IPTG) (Weiss et al., 1999) to induce expression of FtsZ–GFP. To create the strains expressing the GFP fusions and mCherryCYTO, we transformed BS002, BS003, BS004, BS005, BS007 and EC3408 with pKS2. Overnight cultures were back-diluted 1:400 in fresh LB media supplemented with appropriate antibiotics [BS002, BS003, BS004, BS005, BS007: 25 μg ml−1 Ampicillin; EC3408: 15 μg ml−1 Spectinomycin; 60 μg ml−1 Kanamycin], 12 mM L-rhamnose and IPTG (BS007 and BS005, 2.5 μM; BS004 and EC3408, 5 μM; BS002, 50 μM; BS003, 100 μM) (Chen et al., 1999; Ghigo et al., 1999; Weiss et al., 1999). The cultures were incubated at 37°C until an OD600 of ∼ 0.4 was reached.
Strains BS002, BS003, BS004, BS005, BS007 and EC3408 were transformed with pEG4 (7) to produce cells coexpressing ZipA–GFP, FtsA–GFP, GFP–FtsL, GFP–FtsQ, GFP–FtsI or GFP–ZapA and FtsZ-mCherry. Cells were grown and induced essentially as described before (Galli and Gerdes, 2010). Cells were grown in M9 minimal media at 30°C supplemented with 0.2% glucose, 1 μg ml−1 thiamine and 0.1% casamino acids, with the addition of appropriate antibiotics and IPTG (BS007 and BS005, 2.5 μM; BS004 and EC3408, 5 μM; BS002, 50 μM; BS003, 100 μM) (Chen et al., 1999; Ghigo et al., 1999; Weiss et al., 1999). The cultures were incubated until an OD600 of ∼ 0.2 was reached.
A volume of cells corresponding to 0.2 OD600 units was collected from cell cultures. The samples were suspended in loading buffer and resolved by sodium dodecyl sulphate-polyacrylamide gel electrophoresis. Proteins were transferred to nitrocellulose membranes using a semi-dry Transfer-Blot apparatus (Bio-Rad). The membranes were blocked in 5% (w/v) milk and probed with anti-sera to FtsZ (Agrisera, Sweden), FtsI (Weiss et al., 1999), ZipA (Hale and de Boer, 1997) or FtsQ (Dopazo et al., 1992).
Live cell imaging and FRAP measurements
Approximately 6 μl of cell culture was placed on a microscope glass coated with thin pre-made agarose pads [1% (w/v) agarose]. A coverslip was added and the cells were left for ∼ 5 min so that they were immobilized. Confocal imaging and FRAP measurements were performed on a Zeiss LSM700 System (Carl Zeiss, Jena, Germany). Image acquisition time in confocal mode was < 1 s (512 × 512). In FRAP mode the resolution was lowered (256 × 256) in order to gain scanning speed. All FRAP data were analysed in Origin 8.5 (OriginLab Corporation, Northampton). SIM images were acquired on a Zeiss LSM780 ELYRA PS.1 system. For each fluorophore the illumination grids were phase shifted 5 times over the field of view (1024 × 1024) and then rotated 180/5 degrees (i.e. five rotations per image) to generate a data set containing 25 raw images per fluorophore. In general, acquisition time for a single data set was 3–6 s. The SIM images were reconstructed and subsequently analysed in the Zeiss software ZEN2011 black. The calibrated system gave a lateral (xy-) resolution of ∼ 100 nm and a height (z-) resolution of ∼ 275 nm. All other images were analysed using ImageJ (NIH USA). In all imaging and FRAP measurements Cerulean was excited at a wavelength of 405 nm, GFP at 488 nm and mCherry at 555 nm (for SIM imaging mCherry was excited with a 561 nm laser). Appropriate filter settings for the respective fluorophores were chosen in the software (ZEN2011 Black). To reduce bleaching the laser power output was kept below 1% for the 405 nm and 488 nm lasers, and below 5% for the 555/561 nm lasers (total laser power was 5, 10 and 10 mW respectively). Objectives used were either 100× or 63× NA 1.4 oil immersion plan-Apochromat for SIM imaging and 63× NA 1.4 oil immersion plan-Apochromat for confocal imaging and FRAP measurements.
Due to limitations in the ZEN2011 software it was not possible to carry out FRAP and time-lapse imaging simultaneously. Therefore we manually followed cells by acquiring confocal images as they progressed from stage 2 (one dot) into stage 3 (no dot), then immediately carried out the FRAP measurement (the timing between the last acquired confocal image and the FRAP measurement was < 10 s). For bleaching during the FRAP measurements the laser power was set to 100% and applied for ∼ 195 ms. Fluorescence recovery was monitored by acquiring an image every ∼ 190 ms for a total of 6 to 18 s. During the FRAP measurements the green channel was disconnected to gain scanning speed. In the FRAP analysis we followed Skoog et al. (2012), in short; the recovery of fluorescence intensity was recorded within two equally sized circular regions (r ≈ 0.2 μm), one in the middle of the bleached compartment and the other in the middle of the unbleached compartment. To address the fact that the fluorophores were subject to constant photobleaching, the fluorescence recovery data were normalized to the total remaining fluorescence at any given time point t. The normalized fluorescence intensity in the bleached compartment was thus
where F1(t) denotes the fluorescence intensity in the bleached compartment at time t and F2(t) is the fluorescence intensity in the unbleached compartment at that time.
For the cell length measurements, cells were harvested at an OD600 of 0.2 or 0.4 and fixed in the same way as for the immunofluorescence measurements. All strains were imaged under bright field illumination and cell length statistics were acquired using the built in ImageJ plug-in. Cell lengths were determined for more than 100 cells (collected form at least 3 different liquid cultures).
The protocol of Addinall et al. (1996) was followed with minor adjustments. After fixing [2.6% (v/v) paraformaldehyde, 0.04% (v/v) glutaraldehyde, 32.25 mM Na3PO4, pH 7.4] in room temperature for 15 min samples were incubated for 20 min on ice. Cells were washed 3 times in PBS and resuspended in GTE buffer (50 mM glucose, 10 mM EDTA, 20 mM Tris-HCl pH 7.5). The samples were incubated for 1–2 min with lysozyme (10 mg ml−1) and washed twice in PBS. Cells were resuspended in PBS containing 2% BSA and blocking was carried out for 15 min. The cells were washed with PBS and incubated with primary antibodies against FtsZ (Agrisera, Sweden) and GFP (Sigma Aldrich, USA) diluted 1:100 in the same buffer, and then incubated overnight at 4°C. Alexa Fluor 647-conjugated anti-rabbit sera (Molecular Probes) and Oregon Green 488-conjugated anti-mouse sera (Molecular Probes), both at a 1:500 dilution (in PBS containing 2% BSA) were used as secondary antibody against FtsZ and GFP respectively. After incubation the cells were washed at least 5 times in PBS and mounted on glass slides pre-coated with 1% agarose pads. Imaging was performed on a Zeiss LSM700 using appropriate laser and filter settings.
We would like to thank Kenn Gerdes (Newcastle University) for the pEG4 plasmid, Piet de Boer (Case Western Reserve University) for the ZipA anti-sera and Miguel Vicente (Centro National de Biotecnologia) for the FtsQ anti-sera. BS acknowledges support from Kungliga Vetenskapsakademien, Stiftelsen Längmanska Kulturfonden and the Swedish Association for Cytoskeletal Research. The work was supported by grants from the Swedish Foundation for Strategic Research (GvH), the European Research Council (GvH; ERC-2008-AdG 232648), the Swedish Research Council (GvH, DOD) and the US National Institutes of Health (DSW, GM083975). The authors declare that they have no conflict of interest.