ClpXP protease degrades the cytoskeletal protein, FtsZ, and modulates FtsZ polymer dynamics

Abstract

FtsZ is the major cytoskeletal protein in bacteria and a tubulin homologue. It polymerizes and forms a ring where constriction occurs to divide the cell. We found that FtsZ is degraded by E. coli ClpXP, an ATP-dependent protease. In vitro, ClpXP degrades both FtsZ protomers and polymers; however, polymerized FtsZ is degraded more rapidly than the monomer. Deletion analysis shows that the N-terminal domain of ClpX is important for polymer recognition and that the FtsZ C terminus contains a ClpX recognition signal. In vivo, FtsZ is turned over slower in a clpX deletion mutant compared with a WT strain. Overexpression of ClpXP results in increased FtsZ degradation and filamentation of cells. These results suggest that ClpXP may participate in cell division by modulating the equilibrium between free and polymeric FtsZ via degradation of FtsZ filaments and protomers.

Cell division in Escherichia coli requires a precise balance of a multitude of proteins. The machinery that divides the cell into 2 daughter cells is comprised of at least a dozen proteins, most of which are essential (1, 2). The major component involved in this process and the scaffold for assembly of the other cell division components is FtsZ (1, 2). It forms a circumferential ring, the Z-ring, at the site of septum formation where constriction occurs (3).

FtsZ, like its structural homologue eukaryotic tubulin, is a GTPase that assembles into polymers in vitro upon addition of GTP (4). FtsZ polymers are dynamic and subunits are rapidly exchanged (5–7). Positive and negative regulator proteins have been shown to modulate FtsZ polymer dynamics. In E. coli, proteins that promote FtsZ polymerization and/or bundling include FtsA, ZipA, ZapA, and SlmA; conversely, MinC and SulA promote polymer disassembly (2). Two proteins modulating FtsZ assembly in Bacillus subtilis that have homologues in E. coli are Noc, which performs a nucleoid occlusion function similar to SlmA, and ClpX, the ATPase component of the 2-component ClpXP protease (8–10). Studies by Levin and coworkers suggested that B. subtilis ClpX inhibits FtsZ polymerization in vitro by an ATP-independent holding mechanism; however, degradation of FtsZ by ClpXP was not observed in vitro (9, 10).

ClpX is a member of the Clp/Hsp100 class of AAA+ proteins [ATPases Associated with various cellular Activities (11)] and possesses ATP-dependent protein remodeling activities (12). It interacts with a proteolytic partner, ClpP, to form an ATP-dependent protease, ClpXP (12). Structurally, ClpP is composed of 2 stacked heptameric rings with an apical and distal pore and with the active sites on the interior of the cavity formed by the stacked rings (13). ClpX, which forms a hexameric ring, is positioned over one or both ClpP pores and regulates entry into the ClpP cavity (14). ClpX binds substrate polypeptides through peptide recognition motifs located at either the N or C terminus of the substrate. Once bound to a substrate, ClpX uses the energy of ATP hydrolysis to forcibly unfold and translocate the substrate polypeptides through its central pore into the chamber of the bound ClpP protease for degradation (12). A proteomic study to elucidate ClpXP substrates by Baker and coworkers identified FtsZ as one of approximately 50 potential substrates (15). In the following experiments, we show that FtsZ is degraded by ClpXP and suggest that ClpXP can modulate FtsZ polymer dynamics through degradation in E. coli.

Results

ClpXP Modulates FtsZ Polymers Through Degradation in Vitro.

To explore the potential role of ClpXP in FtsZ polymer dynamics, we examined if ClpXP could disassemble and degrade GTP-induced FtsZ polymers in vitro. We labeled FtsZ with a fluorophore and examined labeled polymers by fluorescence microscopy after incubation in the presence or absence of ClpXP. In reactions containing FtsZ and GTP, we detected networks of long thick fluorescent filaments that, on the basis of size, are likely bundled polymers (Fig. 1A). GTP is known to initiate a rapid conformational change in FtsZ inducing polymerization. FtsZ polymers are in a dynamic equilibrium, and subunits within the polymer are rapidly exchanged on a time scale of 7 s (7). As expected, fluorescent polymers were not seen in mixtures that lacked nucleotide (Fig. 1B). Interestingly, when FtsZ was incubated with ClpXP in the presence of GTP and ATP, we detected few FtsZ filaments, suggestive of degradation by ClpXP (Fig. 1C). In a control experiment, we observed that FtsZ polymers were not altered by a combination of ATP and GTP (Fig. 1D), as expected because FtsZ does not bind ATP (16). When mixtures contained ClpXP, FtsZ, and GTP but lacked ATP, filaments were abundant (Fig. 1E), consistent with ClpXP's known specificity for ATP and its inability to use GTP (17).

Fig. 1.

ClpXP degrades FtsZ polymers in vitro. (A–E) Fluorescence microscopy images of fluorescent FtsZ, referred to as FtsZ-488, with and without nucleotides and ClpXP. FtsZ was labeled with Alexa Fluor 488 as described in SI Methods. Reaction mixtures containing 2 μM FtsZ-488 in buffer A supplemented with 25 μg/mL acetate kinase, 15 mM acetyl phosphate, and, where indicated, 4 mM ATP, 2 mM GTP, 2 μM ClpX, and 2.8 μM ClpP were incubated for 90 min at 24 °C, then 12 μL was placed onto a microscope slide, a polyL-lysine-coated coverslip was added, and samples were visualized using a Zeiss LSM 510 NLO fluorescence microscope with a 100× oil objective. Panels show FtsZ-488 with the indicated additions: GTP (A); without nucleotide (B); GTP, ATP and ClpXP (C); GTP and ATP (D); and GTP and ClpXP (E). (F and G) Quantification of polymerized and degraded FtsZ in the presence of ClpX with and without ClpP. In F, [³H]FtsZ was incubated with GTP, ATP, and ClpXP. In G, ClpP was omitted. After incubation, the quantity of FtsZ polymers (filled black circles) and the amount of degraded FtsZ (open red squares) were determined as described in Methods. In F and G, data from 3 replicates are presented as mean ± SEM.

As the visualization of fluorescent polymers by microscopy is not quantitative, we monitored the effect of ClpXP on FtsZ polymerization and degradation by a radioassay. Radioactively labeled FtsZ was incubated with ClpXP, ATP, and GTP. After incubation, the amount of polymerized FtsZ was determined by measuring radioactivity in the pellet fraction following centrifugation, and the amount of FtsZ degraded was determined by measuring acid-soluble radioactivity. With increasing concentration of ClpXP, fewer FtsZ polymers were recovered and more FtsZ was degraded (Fig. 1F). When ClpP was omitted, there was an insignificant reduction in the amount of polymerized FtsZ and no detectable degradation of FtsZ by ClpX alone (Fig. 1G). Together these results suggest that ClpXP degrades FtsZ in vitro. Because FtsZ polymers are dynamic, ClpXP could either be reducing the quantity of FtsZ polymers directly through polymer degradation or indirectly through monomer degradation, which would thereby shift the equilibrium toward depolymerization.

ClpXP Degrades FtsZ Polymers and Monomers.

To address the question of whether ClpXP preferentially degrades FtsZ monomers or polymers, we compared rates of FtsZ degradation by ClpXP in the absence of GTP, the condition favoring monomers, and in the presence of GTP, the condition favoring polymers. Our results show that the rate of degradation of FtsZ by ClpXP, as measured by acid solubilization of radioactive FtsZ, was approximately twofold higher in the presence of GTP (0.46 pmol min⁻¹ ± 0.03) compared with the rate in the absence of GTP (0.26 pmol min⁻¹ ± 0.04; Fig. 2A). In the absence of ClpXP, approximately 60% of the FtsZ was in the polymer form in the presence of GTP and approximately 10% in the absence of GTP, as determined by quantifying the amount of FtsZ associated with the pellet fraction after centrifugation (Fig. 2B). When GTP was substituted with GMPCPP, a slowly hydrolyzed GTP analogue that supports the formation of polymers with slower subunit exchange [Fig. 2B (7, 18)], the rate of degradation was similar to the rate in the presence of GTP (Fig. 2A). In control experiments, there was no detectable degradation in the absence of ATP or in the absence of both ATP and GTP (Fig. 2A). Also, removal of residual GDP, which co-purifies with FtsZ, did not alter the rate of degradation (0.30 pmol min⁻¹ ± 0.03 for apo-FtsZ). We also tested whether the addition of ClpXP directly to pre-formed FtsZ polymers would further increase the rate of degradation. We preincubated FtsZ with GTP for 5 min before adding ClpX, ClpP, and ATP and compared the rate of degradation to the rate measured for a reaction containing all components but with no preincubation. We detected no change in the rate of degradation after preincubation of FtsZ with GTP (Fig. 2C), as expected because assembly of FtsZ polymers occurs rapidly and is dynamic. Therefore, as approximately 60% of FtsZ was in the polymerized conformation with either GTP or GMPCPP and the rates of degradation were faster under these conditions, polymers are likely the preferred substrate for ClpXP, although monomers are also degraded. Degradation rates with and without GTP increased with FtsZ concentration up to 75 μM without reaching a plateau, indicating that high substrate concentration enhances recognition by ClpX (Fig. 2D).

Fig. 2.

ClpXP degrades FtsZ polymers more efficiently than monomers. (A) The rate of degradation of [³H]FtsZ by ClpXP was determined as described in Methods. Where indicated, 4 mM ATP, 2 mM GTP, and 0.5 mM GMPCPP were added. (B) Quantification of [³H]FtsZ polymers collected by centrifugation from reaction mixtures identical to those in A, but without ClpXP. (C) [³H]FtsZ was preincubated with GTP at 24 °C for 5 min, then ClpX, ClpP, and ATP were added. The rate of degradation was compared with the rate measured for a reaction containing all components but with no preincubation step. (D) The rate of degradation of [³H]FtsZ by ClpXP was determined for various concentrations of FtsZ (5–75 μM) with GTP (open red squares) or without GTP (filled black circles). In A-D, data from 3 replicates are presented as mean ± SEM.

To rule out the possibility that GTP stimulates degradation of all substrates by ClpXP, the effect of GTP on degradation of another ClpXP substrate, GFP-SsrA, was tested. We observed that the rate of degradation of GFP-SsrA, as monitored by the decrease in GFP fluorescence with time, was similar in the presence or absence of GTP [supporting information (SI) Fig. S1]. Our results suggest that the stimulation of FtsZ degradation with GTP, or the analogue GMPCPP, is likely because of its effect on the conformation of FtsZ rather than an effect on ClpX.

To substantiate our finding that ClpXP degrades the polymeric form of FtsZ, we wanted to demonstrate that ClpX could interact with FtsZ polymers directly. GTP was added to induce FtsZ polymerization in reaction mixtures containing FtsZ, [³H]ClpX, ClpP, and ATPγS, an ATP analogue that is slowly hydrolyzed by ClpX (14). The amount of ClpX associated with polymerized FtsZ was quantified by measuring radioactivity in the pellet fraction after centrifugation. We detected approximately half of the ClpX in the reaction fractionating with FtsZ when GTP was present (Fig. 3A); when either GTP or FtsZ was omitted, pellet-localized ClpX was reduced by approximately 70%. We varied the ClpXP concentration in the reaction and observed that the amount of ClpX in association with FtsZ polymers increased with increasing concentrations of ClpX (Fig. 3B). These results suggest that ClpX is able to bind FtsZ polymers.

Fig. 3.

ClpX binds to FtsZ polymers and monomers. (A) Quantification of ClpX associated with FtsZ polymers following centrifugation of reaction mixtures containing FtsZ, [³H]ClpX, ClpP, ATPγS, and GTP where indicated as described in Methods. (B) Quantification of complex formation between ClpX and FtsZ polymers as in A, using increasing amounts of [³H]ClpX in reactions containing ClpP and ATPγS. (C) Quantification of [³H]FtsZ monomers bound by ClpX or ClpXP in the presence or absence of ATP as described in Methods. (D) Quantification of ClpX associated FtsZ monomers as in C, using increasing amounts of [³H]FtsZ with ATP and ClpXP (open red squares) or ClpP (filled black circles). In A–D, data from at least 3 replicates are presented as mean ± SEM.

As ClpXP degrades monomeric FtsZ (Fig. 2 A and D), ClpX must also bind to monomers. To monitor complex formation, we incubated ClpX or ClpXP with radioactive FtsZ at 0 °C and measured retention of [³H]FtsZ on Microcon cellulose filters with a 100-kDa exclusion limit (Fig. 3C). We observed specific retention of [³H]FtsZ by ClpX alone and by ClpXP (Fig. 3C) in both the absence and presence of ATP, although less FtsZ was retained when ATP was present (Fig. 3C), likely because of slow ATP hydrolysis resulting in protein unfolding by ClpX and degradation by ClpXP. When ATPγS was included, the results were similar to those with ATP (data not shown). We observed that saturation of ClpXP was obtained with a ratio of ClpX hexamers to FtsZ monomers of approximately 1:1 (Fig. 3D), suggesting it is unlikely that ClpXP binds to more than one FtsZ protomer at a time. In a control experiment, no binding of FtsZ was seen with ClpP alone, as expected (Fig. 3D). Our data suggest that neither nucleotide nor ClpP is required for hexameric ClpX to associate with FtsZ monomers, allowing for the possibility that ClpX may prevent polymerization of FtsZ by an ATP-independent holding mechanism under certain conditions. Contrary to results published for B. subtilis FtsZ and ClpX (9, 10), we have not detected prevention of FtsZ polymerization with E. coli ClpX, either by light scattering experiments or by centrifugation experiments (Fig. S2). Moreover, the presence of ClpP and ATP, as exists in vivo, favors degradation rather than holding.

ClpX N-Terminal Zinc-Binding Domain Participates in Recognition of FtsZ.

ClpX(ΔN61), which lacks the N-terminal 61-aa residues of ClpX, retains its ability to form a complex with ClpP and catalyze degradation of some substrates, such as SsrA-tagged proteins (19, 20). Other substrates, including MuA and λO, are recognized through contact sites in the N-terminal domain and are not degraded by ClpX(ΔN61)P (20, 21). To determine if the N-domain of ClpX is involved in the recognition of FtsZ, we compared degradation of FtsZ by ClpX(ΔN61)P versus degradation by ClpXP (Fig. 4A). Interestingly, the rate of FtsZ degradation by ClpXP in the presence of GTP was approximately 4.3-fold faster than the rate of degradation by ClpX(ΔN61)P, suggesting that the N-domain is important for recognition of FtsZ polymers (Fig. 4A). However, in the absence of GTP, the rate of degradation by WT ClpXP was approximately 1.4-fold faster than ClpX(ΔN61)P, indicating that removal of the N-domain only modestly impairs recognition of FtsZ monomers (Fig. 4B). These results suggest that the N-domain is important for recognition of FtsZ polymers; however, non-polymerized FtsZ is primarily recognized by a ClpX site distal to the N-domain.

Fig. 4.

The N-terminal domain of ClpX and the C-terminal 18 residues of FtsZ are important for substrate recognition. Degradation of [³H]FtsZ as described in Methods using 25 μM FtsZ by either ClpXP (filled red circles) or by ClpX(ΔN61)P (open blue triangles) in the presence (A) or absence (B) of GTP. (C) Light scattering of FtsZ (black) and FtsZ(ΔC18) (blue) polymers. Polymerization was initiated with addition of GTP (1 mM) at 2 min and monitored for 40 min as described in SI Methods. (D) Rates of [³H]FtsZ and [³H]FtsZ(ΔC18) degradation by ClpXP in the presence of GTP as described in A. In A, B, and D, data from 3 replicates are presented as mean ± SEM. In C, a representative experiment of 3 replicates is shown.

The FtsZ C Terminus Is Important for Recognition by ClpX.

ClpXP degrades proteins processively from either the N or C terminus, so we constructed an FtsZ deletion protein lacking the C-terminal 18 aa, FtsZ(ΔC18), to determine if the C terminus engages ClpX. FtsZ(ΔC18) contains an intact polymerization domain (residues 1–320) (22) but lacks the FtsA/ZipA recognition sequence, which is in the C-terminal 18 residues (23, 24). As expected, FtsZ(ΔC18) polymerized to a similar extent as WT FtsZ in light scattering experiments (Fig. 4C). However, FtsZ(ΔC18) was degraded at approximately 15% the rate of WT (Fig. 4D), suggesting that the FtsZ C terminus is required for efficient recognition and degradation by ClpXP. We also tested whether ClpX could interact with FtsZ(ΔC18) polymers and found that there was a twofold reduction in ClpX associated with pelleted FtsZ(ΔC18) polymers compared with full-length FtsZ (Fig. S3).

ClpXP Degrades FtsZ in Vivo.

To address the question of whether FtsZ is degraded by ClpXP in vivo, we compared the half-life of FtsZ in a clpX deletion strain, E. coli MC4100 ΔclpX::kan, versus the WT parental strain by antibiotic chase experiments. Cells were grown to exponential phase (Fig. S4A), spectinomycin was added to block new protein synthesis, and at various times the level of FtsZ in each strain was determined by immunoblot analysis (Fig. 5A). In WT cells the half-life of FtsZ was approximately 115 min or 13% per generation, assuming a generation time of 30 min (Fig. 5B). In the clpX deletion strain, the turnover rate was slower with a calculated half-life of approximately 205 min (Fig. 5B). The slow FtsZ turnover rate in WT cells was expected as FtsZ protein levels have been reported to remain relatively constant with respect to total cell mass throughout the E. coli cell cycle (25).

Fig. 5.

FtsZ is degraded in vivo by ClpXP. (A) In vivo determination of FtsZ stability in MC4100 WT and ΔclpX::kan cells. At various times after addition of spectinomycin, cellular proteins were analyzed by SDS/PAGE and immunoblotting with FtsZ antisera. (B) Half-life of FtsZ in MC4100 WT (black filled circles) and ΔclpX::kan (blue open squares) cells calculated from densitometry analysis of immunoblots in A. (C) In vivo determination of FtsZ stability in MG1655 cells containing control vector (pBR322) or pClpXP as in A. (D) Half-life of FtsZ in MG1655 cells containing pBR322 (black filled circles) or pClpXP (open red squares) calculated from densitometry analysis of immunoblots in C. In A and C, immunoblots shown are representative of 4 independent experiments. In B and D, data from 4 replicate experiments are presented as mean ± SEM.

Next we examined the stability of FtsZ in vivo in the presence of elevated ClpXP protease by comparing FtsZ turnover in E. coli MG1655, carrying either a plasmid overexpressing clpX and clpP from the native promoter, pClpXP, or the vector control plasmid, pBR322 (Fig. 5C), from exponential phase cultures (Fig. S4 B and C). The half-life of FtsZ in cells carrying pClpXP was approximately 45 min (Fig. 5D). In cells carrying the vector alone, FtsZ was more stable, with a half-life of approximately 105 min (Fig. 5D). Therefore, our results from FtsZ turnover experiments suggest that FtsZ is degraded by ClpXP in vivo.

Overexpression of ClpXP Causes Filamentation in E. coli.

Because cells with temperature-sensitive FtsZ mutations exhibit a filamentous phenotype (26) and our studies demonstrate that an increase in ClpXP level causes a decrease in FtsZ level, we examined cellular morphology by phase-contrast microscopy. Populations of MG1655 cells with pClpXP contained normal rod-shaped cells in addition to a significant proportion of cells that were filamentous (28%; Fig. 6A and Table S1). Populations of the WT strain with control vector contained only rod-shaped cells (Fig. 6B). Also, cultures of MC4100 ΔclpX::kan contained rod-shaped cells resembling WT MC4100 cells (Table S1 and Fig. S5), consistent with previous results showing a lack of growth defects in strains defective in ClpX (27).

Fig. 6.

Overexpression of ClpXP induces filamentation. Confocal microscopy images of MG1655 cells containing pClpXP(A) or pBR322 (B). Nucleoids of MG1655 cells containing pClpXP (C) or pBR322 (D) stained with Hoechst dye as described in SI Methods.

Next, we stained cellular DNA with fluorescent Hoechst dye to determine if the filamentous cells overexpressing ClpXP contained evenly spaced nucleoids throughout the length of the filament, as do cells that are incapable of septum formation as a result of mutations in fts genes (2). We observed that the filamentous cells overexpressing ClpXP contained discrete nucleoids and the majority of the nucleoids were evenly spaced along the long axis of the filament (Fig. 6 C and D). These results show that overexpression of ClpXP causes a defect in septum formation, suggesting that FtsZ ring assembly and/or disassembly may be modulated by ClpXP-mediated degradation.

Discussion

Here we describe the involvement of ClpXP protease in the degradation of an E. coli cell division protein. We observed that FtsZ is degraded by ClpXP. Comparison of degradation rates of FtsZ monomers and polymers suggests that ClpXP has higher affinity for FtsZ polymers than monomers. This may be the result of increased exposure of a ClpX recognition signal, likely the C-terminal end of FtsZ, when FtsZ is in the polymerized conformation.

In the current model for the role of FtsZ in cell division (1, 2), the FtsZ-ring forms at mid-cell, providing a scaffold for the assembly of the many factors needed for cell division. The ring then constricts and the 2 daughter cells separate. Following division, the FtsZ-ring disassembles and new FtsZ-rings assemble at the midlines of the new daughter cells (28). In our working model for the mechanism of action of ClpXP in cell division, we suggest that degradation of FtsZ by ClpXP helps to disassemble the FtsZ-ring following cell division and thereby facilitates the recycling of FtsZ molecules for the next round of cell division. Following division, the residual FtsZ filaments in the daughter cells are localized to the new pole. ClpX is able to recognize FtsZ filaments via exposed C-terminal ends of individual subunits and may extract subunits from the filaments by its known mechanism of forcibly unfolding substrates and translocating the polypeptides into the cavity of ClpP for degradation. This process would break the polymer, and repeated cycles of subunit extraction and degradation would result in further disassembly. ClpXP also has the ability to degrade FtsZ monomers, which indirectly could reduce the amount of FtsZ polymers by lowering the local concentration of FtsZ, and consequently shift the equilibrium toward polymer disassembly.

By these possible mechanisms, ClpXP may modulate Z-ring disassembly alongside other cellular factors; such as MinC, which prevents FtsZ polymerization and/or lateral interactions within the polymer (29, 30), and SulA, induced under DNA damaging conditions, which prevents FtsZ assembly and GTP hydrolysis (31, 32). Through the actions of these negative regulators, the FtsZ-ring may be dismantled. FtsZ monomers and oligomers may migrate or diffuse (33) away from the site of the previous cell division toward the midline of the new cell to be recycled in the next round of cell division.

This model accommodates our observation that the overall turnover rate of FtsZ in WT cells is slow (≈110 min; Fig. 5) relative to the length of the cell cycle (≈30 min). Because only 10%–20% of the FtsZ molecules are degraded per generation, the model suggests that ClpXP participates in cell division by modulating monomer-polymer dynamics, rather than reducing the total amount of FtsZ. Our data indicating that overexpression of ClpXP induces filamentation of cells is suggestive that increased degradation of FtsZ slows septum formation; however, we have not ruled out other mechanisms by which ClpXP could induce filamentation, such as degradation of other cell division proteins. The model also incorporates the known facts that there are approximately 100 hexamers of ClpX and approximately 300 tetradecamers of ClpP per E. coli cell (34), whereas there are approximately 10,000 FtsZ molecules per cell (1). As only 1% of the FtsZ could be bound by ClpX, which is known to bind substrates in its hexameric form (12), it is unlikely that E. coli ClpX modulates FtsZ assembly by an ATP-independent holding mechanism as suggested for B. subtilis (9, 10). In addition, there is sufficient ClpP in E. coli such that ClpX exists largely in ClpXP complexes, making it likely that ClpX mostly acts in degradation in conjunction with ClpP. Yet we observed that ClpX interacts with monomeric FtsZ in the absence of nucleotide and ClpP (Fig. 3C), allowing for the possibility that ClpX may prevent polymerization of a small amount of FtsZ under some conditions.

Interestingly, eukaryotic microtubules are remodeled by members of another family of AAA+ proteins, which includes spastin and katanin (35). Spastin recognizes the C-terminal tail of tubulin protomers and promotes ATP-dependent microtubule breaks (36). The mechanism proposed for severing of microtubules by spastin is similar to the mechanism of action of ClpXP acting on FtsZ polymers proposed earlier. Both models predict that the AAA+ hexamer binds to and removes a subunit from within a polymer and proceeds with ATP-catalyzed protein unfolding (37). In the case of ClpXP, the bound protease degrades the subunit, whereas unfolded subunits are likely released by spastin. Both processes illustrate a mechanism by which soluble protein aggregates—in these instances, protein fibers—may be disassembled. Furthermore, this mechanism is similar to the mechanism of protein disaggregation proposed for another Clp/Hsp100 protein, ClpB in bacteria and Hsp104 in yeast, for the disassembly of soluble protein aggregates and amyloid fibers, such as prions in yeast (38).

It is surprising that the roles for ClpX or ClpXP in cell division are widely varied in B. subtilis, E. coli, and Caulobacter crescentus. In C. crescentus, ClpXP acts at the initiation step of cell division by degrading the cell cycle regulator, CtrA, which enables transcription of cell division genes (39, 40). Although FtsZ is degraded in C. crescentus species following cell division, the protease remains unidentified (41). In B. subtilis, ClpX alone can modulate FtsZ assembly and FtsZ fails to be degraded when ClpP is present (9, 10). In E. coli, ClpXP appears to modulate the equilibrium between 2 distinct FtsZ conformations, monomers and polymers, through disassembly and degradation of FtsZ. Although mechanistic details may be dissimilar in different organisms, data suggest that ClpX engages FtsZ and serves as a negative regulator of FtsZ assembly.

Methods

Polymerization Assays.

To measure FtsZ polymer formation by centrifugation, reaction mixtures (25 μL) in buffer A (50 mM Mes, pH 6.5, 100 mM KCl, 10 mM MgCl₂) containing 10 μM [³H]FtsZ, 50 μg/mL acetate kinase, 30 mM acetyl phosphate, and, where indicated, 2 mM GTP, 4 mM ATP, and 0.5 mM GMPCPP (Jena Biosciences), were centrifuged at 259,000 × g for 15 min at 24 °C. Pellet associated radioactivity was determined by scintillation counting. For experiments shown in Fig. 1 F and G, mixtures were assembled as described but with 8 μM [³H]FtsZ, ClpX, and ClpP as indicated. After 10 min at 24 °C, reaction mixtures were centrifuged for 30 min at 259,000 × g and pellet-associated radioactivity was measured as described. Fraction degraded was quantified by measuring trichloroacetic acid (TCA)-soluble (15% wt/vol) radioactivity.

Degradation Assays.

[³H]FtsZ (10 μM) was incubated for 30 min or as indicated in Fig. legends at 24 °C with 1 μM ClpX and 1.3 μM ClpP in reaction mixtures (25 μL) containing buffer A supplemented with 50 μg/mL acetate kinase, 30 mM acetyl phosphate, and 4 mM ATP with 2 mM GTP or 0.5 mM GMPCPP where indicated. Reactions were stopped with 15% (wt/vol) TCA, and TCA-soluble radioactivity was measured. In reactions comparing degradation of FtsZ by ClpXP and ClpX(ΔN61)P, the following protein concentrations were used: 25 μM [³H]FtsZ, 2 μM ClpP, and 1.5 μM ClpX or ClpX(ΔN61).

Protein Binding Assays.

To measure FtsZ polymer-associated ClpX, 0.2 μM [³H]ClpX was added to reaction mixtures (25 μL) containing buffer B (50 mM Hepes, pH 7.2, 100 mM KCl, 10 mM MgCl₂) supplemented with 25 μg/mL acetate kinase, 15 mM acetyl phosphate, 0.5 mg/mL BSA, 1 mM ATPγS, 1.2 μM ClpP, and, where indicated, 10 μM FtsZ and 2 mM GTP. Polymers were collected by centrifugation at 259,000 × g for 15 min at 24 °C.

To measure association of ClpX with FtsZ monomers, 2 μM [³H]FtsZ was incubated at 0 °C for 20 min in reaction mixtures (50 μL) containing buffer A supplemented with 0.01% Triton X-100 (vol/vol), 2.5 μg BSA, and, where indicated, 4 mM ATP, 1.5 μM ClpX, and 2 μM ClpP. Reactions were filtered through Microcon YM-100 cellulose filters (Millipore) by centrifugation at 2,300 × g for 20 min. Retained proteins were recovered with 10% SDS and radioactivity was measured. Background corrections were made by subtracting the amount of [³H]FtsZ retained when both ClpX and ClpP were omitted (14 pmol).

In Vivo Turnover Assay.

Overnight cultures of E. coli MC4100, MC4100 ΔclpX::kan (42), or MG1655 containing either pBR322 or pClpXP were used to inoculate 20 mL of LB with 1.25 × 10⁸ cells with or without carbenicillin (100 μg/mL). At an OD₆₀₀ of 0.7, spectinomycin (200 μg/mL) was added and samples of equal volume were taken at 0, 30, 60, and 120 min. Proteins were precipitated with 15% (wt/vol) TCA. Cell extracts were analyzed by SDS/PAGE and immunoblotting. Densitometry analysis was performed using ImageJ software (National Institutes of Health). Turnover rates were determined in each experiment by measuring the change in FtsZ signal intensity relative to the signal at 0 min.

Additional Methods.

Additional methods may be found in SI Methods.