A Specific Role for the ZipA Protein in Cell Division

Background: ZipA attaches FtsZ to the E. coli inner membrane, its action can be bypassed by FtsA* gain-of-function mutants. Results: FtsZ levels, decreased by ClpP in maxicells, are maintained by an excess of ZipA, but not FtsA+ or FtsA*. Conclusion: ZipA may protect FtsZ from ClpP degradation by preventing recognition by ClpX. Significance: ZipA cannot be fully replaced by the other proto-ring proteins. In Escherichia coli, the cell division protein FtsZ is anchored to the cytoplasmic membrane by the action of the bitopic membrane protein ZipA and the cytoplasmic protein FtsA. Although the presence of both ZipA and FtsA is strictly indispensable for cell division, an FtsA gain-of-function mutant FtsA* (R286W) can bypass the ZipA requirement for cell division. This observation casts doubts on the role of ZipA and its need for cell division. Maxicells are nucleoid-free bacterial cells used as a whole cell in vitro system to probe protein-protein interactions without the need of protein purification. We show that ZipA protects FtsZ from the ClpXP-directed degradation observed in E. coli maxicells and that ZipA-stabilized FtsZ forms membrane-attached spiral-like structures in the bacterial cytoplasm. The overproduction of the FtsZ-binding ZipA domain is sufficient to protect FtsZ from degradation, whereas other C-terminal ZipA partial deletions lacking it are not. Individual overproduction of the proto-ring component FtsA or its gain-of-function mutant FtsA* does not result in FtsZ protection. Overproduction of FtsA or FtsA* together with ZipA does not interfere with the FtsZ protection. Moreover, neither FtsA nor FtsA* protects FtsZ when overproduced together with ZipA mutants lacking the FZB domain. We propose that ZipA protects FtsZ from degradation by ClpP by making the FtsZ site of interaction unavailable to the ClpX moiety of the ClpXP protease. This role cannot be replaced by either FtsA or FtsA*, suggesting a unique function for ZipA in proto-ring stability.

transmembrane segment, and a large cytosolic C-terminal part (1,2). Together with FtsA, it helps to anchor the cytoplasmic FtsZ to the membrane forming a proto-ring at an early phase of divisome formation at midcell (revised in Refs. 3,4). Although an FtsZ-ring can assemble at midcell either in the presence of FtsA or ZipA, its long term stability depends on the simultaneous presence of both proteins (5)(6)(7). The largely flexible cytoplasmic region of ZipA contains three domains: the Map-Tau repeat, the proline-glutamine-enriched region (P/Q domain) 5 and the globular domain at the C terminus. Structural data show that the globular domain (also named FtsZ-binding domain, FZB) interacts directly with FtsZ (8). The ZipA-FtsZ interaction seems to be limited to this region because the P/Q domain most likely folds as a long extended unstructured arm linking the bound FtsZ molecule to the cytoplasmic membrane. No function in FtsZ binding has been assigned to the Map-Tau repeat because topologically it is located near the membrane surface, thus being far away from the FZB domain (2,9). FtsA, a member of the actin/Hsp70/sugarkinase superfamily, is a cytoplasmic protein able to bind ATP (10 -12). The structural features of the region of FtsA-FtsZ interaction have been determined for the crystallized Thermotoga maritima FtsA in the presence of the T. maritima FtsZ C-terminal peptide (13). Whether the binding of the FtsZ-ring to the membrane occurs alternatively via ZipA or FtsA or whether both proteins are needed simultaneously is still unresolved. Several FtsA gain-offunction mutants have been isolated and described to bypass the need for ZipA (14,15). Among the best characterized, FtsA* contains a single amino acid substitution of arginine to tryptophan at position 286 (14). This mutant can bypass the ZipA requirement for assembling a division ring and also suppress the cell division inhibitory effect produced by an excess of ZipA (14). FtsA* has a stronger effect on the stabilization of the FtsZring (14). As the role of ZipA can be partially bypassed by FtsA* and as the zipA gene has only been identified in ␥-proteobacteria (16,17), the role of ZipA in cell division has been proposed to be accessory.
We have studied the ability of different forms of ZipA to stabilize the FtsZ protein in the presence of different amounts of FtsA or FtsA* in E. coli. As a convenient system to analyze it we have used maxicells, which do not contain nucleoids but can sustain expression of individual genes introduced in suitable plasmids (18). The results contribute to define the role of ZipA in bacterial division.

EXPERIMENTAL PROCEDURES
E. coli Maxicells-E. coli CSR603 maxicells were prepared, with minor modifications, as reported by Sancar et al. (18). Briefly, the E. coli strain CSR603 (F-, thr-1, araC14, leuB6(Am), ⌬(gpt-proA)62, lacY1, tsx-33, glnV44(AS), phr-1, galK2(Oc), Ϫ , Rac-0, gyrA98(NalR), recA1, rpsL31(strR), kdgK51, xylA5, mtl-1, argE3(Oc), thi-1, uvrA6) (19) was grown overnight in a shaker at 37°C in LB supplemented with antibiotic-dependentcarrying plasmid (supplemental Table S1). Next day, 20 ml of fresh LB, supplemented with antibiotics if necessary, was inoculated 1:100 with the grown overnight cell culture and incubated in a shaker at 37°C until the cell culture reached a cell number of 1 ϫ 10 8 cells/ml. 10 ml of the bacterial suspension was irradiated in complete darkness with ultraviolet light (UV) (UV dose at 254 nm: 105.98 joules/m 2 ) and incubated for 3 h in a shaker at 37°C in complete darkness to induce the degradation of UV-damaged chromosomal DNA (18). 3 h after UV irradiation, D-cycloserine at a final concentration of 200 g/ml was added to the culture, and it was left for additional 16 h in a shaker at 37°C in complete darkness. D-Cycloserine inhibits the condensation of the pentapeptide peptidoglycan precursors during peptidoglycan synthesis and therefore lyses actively dividing bacterial cells that escaped UV damage (20). The obtained maxicells were processed directly for phase-contrast, immunofluorescence microscopy, SDS-PAGE, or Western blotting. For a further discussion of bacterial cell growth, see supplemental Experimental Procedures.
Plasmids-The plasmids and primers used in this work are listed in supplemental Tables S1 and S2, respectively. Pyrococcus wosei (Pwo) DNA polymerase (Roche Applied Science), and a primer annealing temperature of 55°C was used for standard PCR (21). The obtained PCR products were checked by DNA sequencing (Parque Científico, UAM, Madrid). All cloning and subcloning steps were done in E. coli DH5␣ using standard cloning procedures (21). The ftsA gene of pPNV40 was obtained by PCR using primers PN1 and PN2 and plasmid pMFV12 (22) as a template and cloned into plasmid pTrc99A. pPZV23 was constructed by subcloning the his::zipA containing a NcoI-BamHI fragment of pET-15ZIP (2) into the plasmid pTrc99A. pPZV24 was constructed by the removal of a 477-bp KpnI fragment of plasmid pPZV23. pPZV29, pPZV30, pPZV32, and pPZV36 were constructed by cloning the PCR products obtained using primer MP26 in combination with MP27, MP29, MP28, or MP31, respectively, and pPZV23 as a template. pPZV33 was obtained according the site-directed mutagenesis kit (Stratagene) using primers MP24 and MP25 and pPNV40 as template. pPZV38 was constructed by cloning the PCR product obtained using primers MP32 and MP33 and pPZV23 as template. pPZV43 was constructed by subcloning the his::zipA NcoI-XbaI fragment of pPZV23 into pBAD24. pPZV128 was constructed by subcloning the his::zipA NheI-SalI fragment of pPZV43 into pBAD33. pPZV131 and pPZV132 were con-structed by subcloning the obtained PCR products using primers MP36 and MP37 with template pPNV40 or pPZV33, respectively.
P1 Transduction-Donor strains P1vir lysate preparation of strains JW0427 (clpP::aph) and JW0428 (clpX::aph) (23) was performed according to Miller (24) and transducted into E. coli CSR603. The obtained transductant were screened on LB plates supplemented with 0.2% citrate together with 10 g/ml kanamycin and 30 g/ml nalidixic acid and incubated at 37°C. Positive candidates were checked by PCR as described by Baba et al. using specific set of primers for aph, clpP, and clpX (23).
Microscopy-For phase-contrast and fluorescence microscopy, cells were spotted on an agarose-padded microscope slide and imaged with an Olympus DP70 camera coupled to a widefield motorized Olympus BX61 microscope, equipped with a 100ϫ immersion oil lens, an Olympus HQ:CY3 filter (excitation band, 530 -545 nm; emission band, 610 -675 nm; beam splitter, 565LP) for phase-contrast optics, an Olympus U-MNU2 filter (excitation filter, 360 -370 nm; emission filter, 420 nm; dichromatic filter, 400 nm) to visualize 4Ј,6-diamidino-2-phenylindole (DAPI) fluorescence and an Olympus U-MWIY2 filter (excitation filter, 545-580 nm; emission filter, 610 nm; dichromatic filter, 600 nm) to visualize the Alexa Fluor 594-labeled rabbit IgG (Invitrogen). The images were captured with the analySIS imaging software (Olympus) and further processed with Adobe Photoshop CS and the Huygens professional deconvolution software package (Scientific Volume Imaging bv, Hilversum, The Netherlands). Three-dimensional reconstruction are composed of individual optical sections (n ϭ 55) that were deconvoluted using the Huygens Professional software package and reconstructed with the Bitplane Imaris Software package (Bitplane, Saint Paul, MN). The elongation along the z axis introduced by the optical system was corrected in the deconvoluted stacks by multiplying the vertical scale by 0.42 (the correction factor calculated from the measurement of control spherical particles).
Immunolabeling-Cells for immunolabeling were prepared as described by Addinall et al. (5), using primary antibodies MVC2, MVM1, or polyclonal antiserum MVC1 to specifically detect FtsZ, FtsA, and ZipA, respectively, and the fluorescent Alexa Fluor 594-labeled anti-rabbit IgG, as secondary antibody. The coverslip was mounted together with the addition of Vectashield Mounting medium (Vector Laboratories) and DAPI on a microscope slide and sealed with transparent nail polish to protect sample from drying during microscopy.
Other Techniques-SDS-PAGE (25) and Western blot analysis (26) were performed according to standard procedures. For SDS-PAGE, the pelleted bacterial cells were resuspended in SDS-PAGE sample buffer and boiled for 5 min at 95°C. Unless stated otherwise, cells corresponding to A 600 of 0.1 were loaded per lane on a SDS-polyacrylamide gel. For Western blot analysis, the polyclonal antibodies MVC2 (FtsZ) and MVM1 (FtsA) (11), and polyclonal antiserum MVC1 (ZipA) (27) were used. Monoclonal Anti-polyhistidine Peroxidase Conjugate Clone HIS-1 (Sigma-Aldrich) was used for detection of histidinetagged ZipA and its mutant variants. Horseradish peroxidasecoupled protein A (Bio-Rad), the BM chemiluminescence blotting substrate (POD; Roche Applied Science), and Kodak Biomax XAR films (Sigma) were used to develop the PVDF (Millipore) or Nitrocellulose (Roche Applied Science) membrane-blotted proteins. Quantification of the immunoblotted protein levels was performed using ImageJ software package (Wayne Rasband, NIH). Measurements of protein contents were done using biological triplicates for each point, and the mean Ϯ S.D. bars are indicated in the graphs. Structural alignment of the C-terminal domains of E. coli (PDB 1F47 (8)) and T. maritima (PDB 4A2A (13)) were performed according to Szwedziak et al. (13) and imaged with MacPyMOL software (The PyMOL Molecular Graphics System, version 1.5.0.4, Schrödinger, LLC) and further processed with Adobe Photoshop CS.

RESULTS
Stability of Proto-ring Components-To study the formation of the proto-ring in vitro we used E. coli maxicells, chromosome-less bacterial cells that retain complex cellular activities as protein synthesis and protein targeting (18,28). Maxicells were obtained by low dose UV irradiation of a recA mutant strain to induce chromosome degradation (18). We first determined the levels of the wild type FtsZ, FtsA, and ZipA proteins at the different stages during maxicell production. No change in the levels of FtsA or ZipA was observed at any stage (Fig. 1). The FtsZ levels also remained largely unchanged at 3 h after UV irradiation relative to those in the non-UV-irradiated cells, but they dropped to a low residual amount in maxicells (Fig. 1).
In growing E. coli cells the ClpXP protease complex has been reported to be responsible for FtsZ degradation (29,30). Protein degradation by ClpXP is directed by the presence of a ClpX recognition signal on the target protein (31). Although the exact recognition motif on FtsZ needs to be identified, Camberg et al. limited its position to the last 18 amino acids of the FtsZ C terminus (29). Upon ATP hydrolysis, the ClpX moiety unfolds and transfers FtsZ to ClpP, the protease moiety (32), where it will be degraded (29). FtsA or ZipA has not shown to possess a ClpX recognition motif. To check whether FtsZ degradation in maxicells depends on the action of ClpXP, strains VIP978 (clpP::aph) and VIP979 (clpX::aph) were constructed by insertional inactivation of clpX or clpP in strain CSR603 (Fig. 1). Maxicells obtained from VIP978 showed no degradation of FtsZ relative to the non-UV-irradiated control, indicating that the protease moiety ClpP is required for degradation. Surprisingly, the absence of ClpX in VIP979 had no effect on preventing the FtsZ degradation in VIP979 maxicells. We cannot exclude that a different recognition unfoldase, as ClpA (33)(34)(35), may take over the role of ClpX in maxicells. Furthermore, ClpP has been reported to be able to produce some limited proteolysis independently of ClpX or ClpA (36,37) although the detailed molecular mechanism is not known.
ZipA Protects FtsZ from Degradation-We have checked the effect of the two other proto-ring components, FtsA and ZipA, on the stability of FtsZ. For this purpose we have induced the production of different forms of ZipA at the time of D-cycloserine addition and measured their effect on the amount of FtsZ present after 16 h in maxicells. Results shown in Fig. 2 and supplemental Fig. S1 indicate that the overproduction of His-ZipA reduced FtsZ degradation. Production of a His-ZipA3 mutant, a C-terminal truncate of His-ZipA that lacks most of its FZB domain (from amino acids 114 to 272, Fig. 3A), showed no stabilization of the FtsZ levels because it also occurred in maxicells carrying an empty vector as a control ( Fig. 2 and supplemental Fig. S1). The fact that maxicells retain a basal level of ZipA, even when they contain no plasmid-encoded copy of zipA, could contribute to maintain the low residual amount of FtsZ found after the D-cycloserine treatment. The FZB domain has been described to interact directly with the C terminus of FtsZ (8); this is the same FtsZ domain recognized by ClpX (29). ClpX is an unfoldase that facilitates the access of ClpP to the substrate, although other unfoldases such as ClpA may be also active. Our results could be due to an steric hin-   FEBRUARY 1, 2013 • VOLUME 288 • NUMBER 5 drance from the excess of ZipA blocking the access of ClpX to the recognition signal present at the C terminus of FtsZ. In the presence of ZipA the accessibility of the ClpP protease to FtsZ would be reduced, and degradation would then proceed at a lower rate. Nevertheless, FtsZ would never be fully protected from degradation by ClpP because the other unfoldases together with the intrinsic protease activity of ClpP may remain active.

FtsZ Stabilization by ZipA
Protection of FtsZ by ZipA Only Requires the FZB Domain-To find which subdomains of ZipA are involved in FtsZ protection, different ZipA deletion mutants containing an intact N-terminal transmembrane domain were constructed as shown in Fig. 3A. In addition, the isolated cytoplasmic FZB domain that has been described to interact with FtsZ (8) was also constructed (Fig. 3A). Overproduction in maxicells shows that only the full-length His-ZipA, the entire FZB domain, and the His-ZipA4 mutant (lacking part of the FZB domain from Pro-273 to Ala-328) were able to protect FtsZ from degradation ( Fig. 3B and supplemental Fig. S2). This result indicates that the isolated FZB domain, or even a shorter form in which the last 55 C-terminal amino acids are missing, is sufficient for FtsZ protection from proteolysis and that the rest of the ZipA protein is not involved in it.
Neither FtsA nor FtsA* Can Protect FtsZ from Proteolysis-The essentiality of the role of ZipA in E. coli division has been questioned because it can be bypassed by gain-of-function FtsA mutants. Among them, FtsA* has been described to allow the late divisome assembly steps in the absence of ZipA (14,38). To test whether FtsA or FtsA* can exert the protective role on the stability of FtsZ that we have observed, we have overproduced FtsA or FtsA* in maxicells. The ftsA or ftsA* gene was placed under the control of an IPTG-inducible P trc promoter (pPNV40 and pPZV33, respectively), and expression was induced at 3 h after UV irradiation at the time of the D-cycloserine addition. The results in Fig. 2 and supplemental Fig. S1 show that when FtsA or FtsA* was overproduced the FtsZ levels diminished in a way similar to that in the control. This indicates that, contrary to ZipA, neither FtsA nor FtsA* protects FtsZ from degradation. Similarly to ZipA, FtsA interacts with the C terminus of FtsZ (13,39), and this interaction was postulated to stabilize and tether the FtsZ-ring to the inner membrane (40). Moreover, a decrease of 5% in the FtsZ protein levels was found in a exponentially growing culture of a ftsA* ⌬zipA strain (WM1657) relative to the wild type cells (WM1074) growing under similar conditions (14) (supplemental Table S3). Our results suggest that stabilization of the ring by FtsA may be exerted at the stages from FtsZ polymerization to ring assembly, but it is very unlikely that it takes place on monomeric FtsZ. We therefore conclude that ZipA has a role on the stability of the FtsZ protein itself and that this role cannot be replaced by either FtsA or FtsA*.
FtsA or FtsA* Does Not Interfere with ZipA in the Protection of FtsZ-Because both ZipA and FtsA can interact with the same region of FtsZ, namely the C terminus, we have tested whether the protective role of ZipA on FtsZ stability could be hindered by the presence of FtsA or even more by FtsA* (41). We overproduced ZipA (pPZV128, arabinose-induced) together with either FtsA (pPNV40, IPTG-induced) or FtsA* (pPZV33, IPTG-induced) in maxicells, inducing expression at the time of D-cycloserine addition, and measured the levels of FtsZ after 16 h. In these conditions, overproduction of FtsA or FtsA* together with ZipA had no effect on decreasing the FtsZ levels significantly relative to those attained when ZipA was produced in isolation ( Fig. 4 and supplemental Fig. S3). We conclude that the presence of FtsA or FtsA* had no effect on either increasing or diminishing the protective action exerted by ZipA on FtsZ stability. To exclude that FtsA and FtsA* may lose their activity if they were mislocalized at inclusion bodies under overproduction conditions, we checked their localization inside the maxicells. Under these experimental conditions the overproduced FtsA and FtsA* proteins localized along the cell membrane, suggesting that their lack of action on ZipA-mediated protection of FtsZ is not an artifact caused by the formation of aggregates (supplemental Fig. S4).
As expected, neither FtsA nor FtsA* was able to exert any protective action on FtsZ when they were combined with the His-ZipA1 protein, itself unable to protect FtsZ from degradation ( Fig. 5 and supplemental Fig. S5). Equally, the protective effect of the His-ZipA4 and FZB proteins on FtsZ was not decreased or increased by the simultaneous presence of either FtsA or FtsA*. These results further support our conclusion indicating that in maxicells the ZipA FZB domain can by itself protect FtsZ from degradation and that it cannot be replaced by either FtsA or FtsA*.
ZipA Overproduction Promotes Ordered FtsZ Structures in Maxicells-Biochemical data have previously shown that ZipA is able to promote FtsZ polymerization, bundling, and sheet formation in vitro (2). Immunofluorescence microscopy images of maxicells show that FtsZ localized from pole-to-pole as regularly spaced foci when His-ZipA was overproduced, whereas no foci were detected in the non-ZipA-overproducing control (supplemental Fig. S6). A three-dimensional reconstruction of deconvoluted FtsZ images identified the regular spaced FtsZ foci as being part of a wide helical structure within the maxicell (Fig. 6).
As expected, the ZipA contained in maxicells was localized along the cytoplasmic membrane (6). Besides lining the membrane, ZipA accumulated at several discrete points along the membrane length when overproduced (supplemental Fig. S4). This local accumulation of overproduced ZipA was more evident when observing the three-dimensional reconstruction of deconvoluted images ( Fig. 6 and supplemental Fig. S4). Although ZipA may have an effect on the topology of FtsZ assemblies, we have not investigated further whether the ZipA foci may coincide with specific regions within the FtsZ helix. ZipA was able to maintain the localization pattern of FtsZ and its own even in the presence of overproduced FtsA or FtsA*  (supplemental Fig. S4). No FtsZ helical spirals were observed when FtsA, FtsA* (supplemental Fig. S6), or His-ZipA3 (data not shown) was overproduced, which is in agreement with their inability to protect FtsZ from degradation.

DISCUSSION
ZipA is an important component of the E. coli division machinery (3). Together with FtsZ and FtsA, it forms the protoring, a structure regarded as the first stage in the assembly of the cytoplasmic division ring (4). In this work we have addressed the effect of ZipA or FtsA on the stability of FtsZ. Our results in Fig. 1 show that in the nucleoid-free maxicells (18) FtsZ is present at low levels and that this is likely due to both the specific degradation of FtsZ by ClpXP and to the absence of de novo synthesis.
Whereas overproduction of ZipA reduces the degradation of FtsZ in maxicells, the overproduction of the other proto-ring component FtsA ϩ or the gain-of-function mutant FtsA* (41) does not. Because all but one of the 18 last FtsZ C-terminal amino acid residues (Ala-366) that are needed for its recognition by ClpX (29) coincide with the sequence that interacts with ZipA (8), we propose that masking of the of FtsZ by the ZipA FZB domain underlies the observed protection. Although FtsA, and presumably FtsA*, also binds to the same C terminus of FtsZ as ZipA, the nature of each interaction is different (Fig.  7A). The FtsZ C terminus is embedded into a hydrophobic pocket of the ZipA FZB and is folded as an extended ␣-helix, exposing its hydrophobic residues at one side of the helix toward the binding interface (8). On the other hand, the binding of the C-terminal ␣-helix of FtsZ to FtsA is electrostatic, and the hydrophobic residues are buried inside the bent allowing the peptide chain to create a water-bound flat polar interface with FtsA (13). These differences in binding may be invoked to explain the different behavior of ZipA or FtsA in FtsZ protection, but the specific details of each interaction need to be investigated further.
We also observe that the protection of FtsZ by ZipA is not affected by the simultaneous presence of either FtsA or FtsA* (Fig. 4). Detailed examination of the structures adopted by the ZipA FZB or FtsA when interacting with FtsZ suggests that due to steric impediments it is not possible for both proteins to bind simultaneously to the same FtsZ molecule (Fig. 7B).
FtsA* has been shown to bypass some of the functions exerted by ZipA in the assembly and stability of the divisome, among them the need for ZipA in the assembly of late divisome proteins (14,38,41), and therefore we expected that it could act by protecting FtsZ from degradation. Instead, we found that neither FtsA nor FtsA* protects FtsZ in maxicells. It is possible then that the effects of ZipA take place on the stability of the FtsZ protein, whereas those of FtsA* are exerted on the stability of the FtsZ ring rather than on the protein itself. Similar to our observations on FtsZ ϩ , whereas overproduction of ZipA has been shown to stabilize the rings formed by FtsZ84 at the nonpermissive temperature (2), FtsA* has no effect on the stability of the FtsZ84 rings at 42°C (41). These published results did not include data on the effects of either ZipA or FtsA* on the protection of FtsZ84.
In addition to the interactions with FtsA, ZipA, and ClpX, the C terminus of FtsZ also interacts with the C-terminal domain of MinC, the FtsZ polymerization inhibitor of the Min system (42). Depending on which molecule is accepted for the interaction, the fate of the FtsZ molecule will be directed to the production of a division ring or prevented from acting in cytokinesis (Fig. 8). In Bacillus subtilis SepF and EzrA, positive and negative modulators of the FtsZ polymerization, respectively, also bind to FtsZ at this site (43)(44)(45). This suggests a role for the C terminus end of FtsZ as a central hub to integrate different   signals that modulate the efficiency and progress of divisome assembly and for ZipA as having a dual role in the anchoring and the stability of FtsZ.