Bacterial SOS Checkpoint Protein SulA Inhibits Polymerization of Purified FtsZ Cell Division Protein

Abstract

Cell division of Escherichia coli is inhibited when the SulA protein is induced in response to DNA damage as part of the SOS checkpoint control system. The SulA protein interacts with the tubulin-like FtsZ division protein. We investigated the effects of purified SulA upon FtsZ. SulA protein inhibits the polymerization and the GTPase activity of FtsZ, while point mutant SulA proteins show little effect on either of these FtsZ activities. SulA did not inhibit the polymerization of purified FtsZ2 mutant protein, which was originally isolated as insensitive to SulA. These studies define polymerization assays for FtsZ which respond to an authentic cellular regulator. The observations presented here support the notion that polymerization of FtsZ is central to its cellular role and that direct, reversible inhibition of FtsZ polymerization by SulA may account for division inhibition.

The bacterial SOS response to DNA damage represents one of the first examples of what has become known as checkpoint regulation (for a recent review, see reference 40). The sulA gene was first identified as a factor required for SOS-mediated division inhibition (16) and was found independently to be a suppressor of lon mutants (15). Transcription of the sulA gene is induced as part of the SOS response in Escherichia coli cells (21, 28). Elevated levels of the SulA division inhibitor are sufficient to cause division inhibition (22).

The target of SulA is thought to be the FtsZ protein (3, 19, 27). FtsZ plays a central role in bacterial division, forming a structural element at the division site (for reviews, see references 6, 11, 26, and 36). Although the amino acid sequence of FtsZ includes only a short segment with a high degree of sequence homology to tubulins, there is a remarkable similarity between the structures of these two proteins (25, 31). Purified FtsZ has GTPase activity (9, 29, 34) and can polymerize to form protofilaments which closely resemble the protofilaments in microtubules (12). FtsZ normally becomes localized to the division site as one of the earliest detected events in cell division (5). However, when SulA is induced, FtsZ fails to localize to the division site (3).

Several lines of evidence point to a direct interaction between SulA and FtsZ. FtsZ stabilizes SulA in vivo (23), and purified SulA binds to FtsZ in vitro (19). Also, wild-type SulA interacts with FtsZ in the yeast two-hybrid system, while mutant SulA proteins do not. Most SulA-resistant FtsZ point mutations abolish interaction with SulA in the two-hybrid system (20).

Diverse conditions have been found to support polymerization of FtsZ in vitro; however, most of these studies were performed either at or below neutral pH (7, 12, 30) or in the presence of calcium (42) at levels significantly higher than that normally found in the cell (8, 13).

We devised conditions that allowed quantitation of FtsZ polymerization in the presence of DEAE-dextran and used these assays to test the effect of SulA on polymerization. The results presented in this report indicate that SulA binding to FtsZ directly inhibits the polymerization of FtsZ.

MATERIALS AND METHODS

Reagents.

Chemicals and reagents were obtained from Sigma unless otherwise stated. DEAE-dextran (Sigma 9885; average molecular weight of 500,000). GTP and deoxynucleoside triphosphates for PCR reactions were purchased from Boehringer, [α-³²P]GTP was obtained from Dupont-NEN, Tris buffer (pH 7.4) was from BioWhittaker, IPTG (isopropyl-β-d-thiogalactopyranoside) was from National Laboratory Source, Syto-17 came from Molecular Probes, and Permount was obtained from Fisher. Restriction enzymes were purchased from New England Biolabs; the DNA polymerases for PCR and the QuickChange site-directed mutagenesis kit were from Stratagene.

Strains and plasmids.

E. coli strains BL21(DE3) [F⁻ ompT hsdS_B (r_B⁻ m_B⁻) gal dcm (DE3)] and HMS174(DE3) [F⁻ recA1 hsdR(r_K12⁻ m_K12⁺) Rif^r (DE3)] (39) were obtained from Novagen and were used as expression hosts for genes cloned under T7 promoter control (the lambda prophage carries the T7 RNA polymerase expressed from the lacUV5 promoter). The recombination-deficient strain DH5alpha (recA1) (18), which lacks any T7 RNA polymerase, was used as the initial cloning host for all subcloning and mutagenesis studies.

Plasmids were obtained from Novagen. The pBR322-derived pET11a (10) confers ampicillin resistance and allows cloning of genes downstream of the T7 promoter overlapped by the lac operator to provide regulation by IPTG, while the pACYC184-derived pLysS plasmid (38) confers chloramphenicol resistance and provides a low level of the T7 lysozyme, which inhibits the T7 RNA polymerase stoichiometrically, thereby reducing the expression of genes cloned in pET11 vectors prior to induction with IPTG (used for PrtA-SulA expression). SulA fusions were constructed in a derivative of pET11a into which a portion of the Staphylococcus protein A gene was inserted, allowing in-frame fusion to the C terminus of protein A for any protein initiating with a methionine codon overlapping an NdeI site (19a). Mutant derivatives of PrtA-SulA and FtsZ were constructed with the QuickChange site-directed mutagenesis kit as described by the manufacturer, starting with the appropriate pET11 clone. Primer pairs used to introduce the mutations were 5′-GGTCAGCAATCGCACTGGCAACTCTGG-3′ and 5′-CCAGAGTTGCCAGTGCGATTGCTGACC-3′ for SulA62H and 5′-GGTCAGCAATCGAGCTGGCAACTCTGG-3′ and 5′-CCAGAGTTGCCAGCTCGATTGCTGACC-3′ for SulA62S. To change FtsZwt to FtsZ2, the primers 5′-CGTGGACTTTGCAGGCGTACGCACCGTAATG-3′ and 5′-CATTACGGTGCGTACGCCTGCAAAGTCCACG-3′ were used. Each clone and mutant derivative were confirmed by using fluorescent dideoxy terminator sequencing on an ABI377 DNA sequencing machine.

Purification of protein A-SulA fusion proteins.

Strain HMS174(DE3)(pLysS) was transformed with a pET11 derivative encoding the fusion to protein A of wild-type SulA or mutant SulA. Cells were grown in minimal medium (33) to an optical density at 600 nm (OD₆₀₀) of 0.5, induced by the addition of IPTG to a final concentration of 1 mM, and grown at 37°C for 2.5 h (for PrtA-SulAwt and PrtA-SulA62H) or at 22°C overnight (for PrtA-SulA62S). Cells were harvested and resuspended in buffer A (10 mM Tris-Cl, pH 8; 10% glycerol; 1 mM dithiothreitol; 1 mM EDTA, 150 mM KCl) to which lysozyme was added to a final concentration of 200 μg/ml. Cells were frozen in liquid nitrogen, thawed, and centrifuged at 40,000 rpm in a 45 Ti rotor for 1 h at 4°C to remove debris. To this extract solid ammonium sulfate was added (0.25 g/ml of extract) to precipitate the PrtA-SulA fusion. Ammonium sulfate pellets were redissolved in buffer B (20 mM Tris-Cl, pH 7.4; 1 mM dithiothreitol; 20% glycerol), conductivity was adjusted to be equivalent to 125 mM NaCl, and then the samples were loaded onto a Q Sepharose fast-flow column (Pharmacia). Protein was eluted by applying a gradient of 125 to 500 mM NaCl in buffer B. Fractions containing the fusion protein were pooled and loaded onto an immunoglobulin G (IgG)-Sepharose column (Pharmacia) equilibrated with TST buffer (50 mM Tris-Cl, pH 7.6; 150 mM NaCl; 0.05% Tween 20). The column was washed with 10 bed volumes of TST buffer then with 2 bed volumes of 5 mM ammonium acetate (pH 5.0). PrtA-containing fractions were eluted with 0.1 M glycine-HCl (pH 3), and fractions were neutralized immediately following elution by the addition of a predetermined volume of 1 M Tris base and then verified to be at a pH of 7.4.

FtsZ polymerization assays.

Wild-type and mutant FtsZ protein was spun at 22,260 × g for 20 min at 4°C in a TLA-100.2 rotor in a Beckman TL-100 ultracentrifuge immediately prior to reaction assembly to remove any preexisting polymers. Reactions were assembled on ice and contained 100 mM Tris-Cl (pH 7.4), 1.5 mM magnesium acetate, 50 mM KCl, 1 mM GTP, and 2 μM FtsZwt or FtsZ2 in a final volume of 200 μl. Reactions were initiated by the addition of DEAE-dextran, incubated for 10 min at 37°C, and then spun at 22,260 × g for 20 min at 37°C. Pellets were resuspended in 200 μl of 100 mM Tris-Cl (pH 7.4). To both supernatant and resuspended pellets, 20 μl of 4× sodium dodecyl sulfate (SDS) sample buffer (0.25 M Tris-Cl, pH 6.8; 8% SDS; 40% glycerol; 20% 2-mercaptoethanol) was added and then the mixture was heated at 95°C, uncovered, for 10 min. Aliquots were loaded onto 10% polyacrylamide Tris-glycine-SDS resolving gels (Bio-Rad). Gels were Coomassie blue stained, destained, scanned with a Molecular Dynamics densitometer, and quantified by using ImageQuant software. Dilution series of known amounts of FtsZ in the gels were used as internal standards.

Affinity copurification of PrtA-SulA and FtsZ.

FtsZ proteins were centrifuged prior to reaction assembly as described above. Reactions were assembled exactly as for the polymerization assays, except that DEAE-dextran was omitted. Where indicated, FtsZ was present at 4 μM and PrtA-SulA fusions were present at 2 μM. After 15 min at 37°C, 200-μl reaction mixtures were loaded onto IgG columns (0.5-ml bed volume), equilibrated in reaction buffer plus 1 mM GTP, washed in the same buffer, and eluted as described above for the PrtA-SulA fusions. Electrophoresis and densitometric analysis was done as described for the sedimentation assay (except that dilutions of PrtA-SulA standards were used in estimating PrtA-SulA).

GTPase assay.

FtsZ proteins were centrifuged prior to reaction assembly as described above for the polymerization assays. The assays (50-μl final volume) contained 100 mM Tris-Cl (pH 7.4), 1.5 mM magnesium acetate, 50 mM KCl, 0.075 μCi of [α-³²P]GTP, and 0.1 mM GTP. Reactions were usually initiated by the addition of 2 μM FtsZwt or FtsZ2. However, when the effects of PrtA-SulA fusion proteins were being monitored, a 5-min preincubation of the FtsZ and PrtA-SulA fusion was carried out in the absence of GTP, and the reactions were then initiated by addition of GTP. Incubation was carried out in a 96-well plate from which 3.5-μl aliquots were withdrawn at specific time points and transferred to a second plate containing 3.5 μl per well of a quenching mix (2 mM GDP, 0.2% SDS, 20 mM EDTA, 30% glycerol), allowing time point data to be obtained at 15-s intervals (typically at 0, 0.25, 0.5, 0.75, 1, 1.25, 1.5, 1.75, 2, 4, 6, and 20 min). From each quenched time point, a 1-μl sample was spotted onto PEI cellulose F thin-layer chromatography plastic sheets (EM Science); each sheet was dried, then washed for 10 min in methanol, and then dried again and developed in 1 M formic acid and 0.5 M LiCl. The sheets were exposed in a phosphor screen cassette and quantitated with a Storm 860 phosphorimager (Molecular Dynamics). The results are expressed as the total amount of GTP converted to GDP in a 50-μl reaction.

RESULTS

FtsZ polymerization.

Purified FtsZ protein can be induced to form extensive polymers at a pH of 6.0 or lower, and DEAE-dextran has been used at pH 6.5 to induce FtsZ polymer formation (12). Limited polymerization has been observed at or above neutral pH by electron microscopy (12, 30). We tested whether DEAE-dextran will also induce polymerization of FtsZ at pH 7.4, a more physiological pH for cells of E. coli. We used a sedimentation assay to monitor the extent of polymer formation. Figure 1A shows the effect of titrating DEAE-dextran into an FtsZ solution. At approximately equal mass ratio of FtsZ to DEAE-dextran there is efficient recovery of FtsZ in the pellet. The level of polymers remained essentially constant from 5 min to at least 20 min when the reactions contained 1 mM GTP, though after longer times reduced polymerization was seen correlating with depleted GTP levels (data not shown). The polymerization of FtsZ dependent on DEAE-dextran is inefficient in the absence of GTP (20 to 45% of full reaction) or magnesium (30 to 50%), but when GDP is present in place of GTP, polymerization is 30 to 60% of that seen with GTP.

FIG. 1.

Titration of DEAE-dextran into FtsZ sedimentation assay. (A) SDS-polyacrylamide gel analysis of the supernatant and pelleted FtsZ protein with increasing levels of DEAE (final concentrations of 0, 3.125, 6.25, 12.5, 25, 50, 100, 200, and 400 μg/ml, respectively, in lanes 1 to 9). The gel lanes are arranged so that the corresponding supernatant and pellet at a specific level of DEAE are vertically aligned. (B) Same as in panel A, but with FtsZ2 mutant protein in place of wild type. (C) Densitometric scans of the gels in the top two panels were used to determine the fraction of FtsZ sedimented and then plotted as a function of the amount of DEAE-dextran added as described in Materials and Methods.

FtsZ2 characterization.

The FtsZ2 mutant was originally isolated in a screen seeking ftsZ alleles that conferred resistance to the SulA division inhibitor (2). The mutation was introduced into the pET11-FtsZ expression plasmid by site-directed mutagenesis. By using the pET11-FtsZ2 plasmid in the expression host BL21(DE3), the FtsZ2 protein was overexpressed and purified for comparison with the wild-type FtsZ. Its purification was made possible by a procedure identical to that used for the wild-type FtsZ protein (7).

Polymerization of FtsZ2 and wild type was compared at different DEAE-dextran concentrations (Fig. 1B). DEAE-dextran was essential to induce significant polymerization of FtsZ2 under these conditions. However, FtsZ2 required significantly lower concentrations of DEAE to induce polymerization than did wild-type FtsZ protein (Fig. 1C).

The ability of FtsZ2 to hydrolyze GTP was compared with that of wild-type FtsZ at the permissive and restrictive temperatures for ftsZ2 mutant growth. The GTPase activity of FtsZ2 was at least 200-fold lower than that of wild-type FtsZ measured at 30°C and even less active at 42°C (Fig. 2). The lower panel (Fig. 2B; note the very different x and y scales compared to Fig. 2A) shows that FtsZ2 did exhibit GTPase activity above background. The FtsZ2 GTPase activity was reduced at 42°C, at which temperature ftsZ2 mutants are not viable (2).

FIG. 2.

(A) GTPase of FtsZ2 compared with wild-type FtsZ at 30 and 42°C. (B) Replotted detail (note axis changes) of FtsZ2 at 30 and 42°C, extending the time to 60 min. Each FtsZ protein (Zwt or Z2) was present as 2 μM, and the assay was performed as described in Materials and Methods.

Polymer structures.

The nature of the polymers formed by FtsZ and FtsZ2 under these conditions was investigated by electron microscopy of negatively stained samples (Fig. 3). The predominant form of FtsZ polymers is a tube with a diameter of about 20 nm. Sheets and minirings of FtsZ resembling those reported previously (12) were also seen. The relationship between these structures is suggested by the appearance of numerous tubules, which seem to be either partially disrupted or else incompletely formed. These structures seem to exhibit a helical arrangement of the protofilaments in the tubules. A plausible explanation of these observed images is that there is a tubule formed from a pair of FtsZ protofilaments (most likely parallel to one another, as in the sheets) that are wrapped helically, as indicated by the diagram in Fig. 3 (inset). The structures observed for FtsZ2 mutant protein were indistinguishable from those formed by wild-type FtsZ, although FtsZ2 showed somewhat fewer sheets than did wild-type FtsZ. Though less abundant, the polymers of FtsZ formed in the absence of GTP or magnesium or when GDP replaced GTP resembled the various forms seen with GTP (data not shown).

FIG. 3.

Electron micrographs of FtsZ polymers induced by DEAE-dextran, each at a magnification of ×59,000. (A to D) Wild-type FtsZ, illustrating tubules (A), protofilaments aligned vertically (B), and minirings (C and D, arrowheads). (E and F) FtsZ2 showing miniring (E) and tubules (F), with inset diagram of proposed tubule structure represented as one light and one dark protofilament. Polymerization reactions were done as described in Materials and Methods with FtsZ proteins present at 5 μM. Scale bar, 100 nm.

SulA fusions and mutants.

To determine whether our polymerization assay at pH 7.4 reflects the cellular conditions, we decided to investigate the effects of SulA protein on the activity of FtsZ. Initial attempts to isolate SulA protein were hampered by its poor solubility. We found that a fusion of part of the Staphylococcus protein A (PrtA) to the amino terminus of SulA resulted in a soluble PrtA-SulA fusion that was easily purified by IgG-affinity column chromatography. Two null mutant forms of SulA were also made as PrtA-SulA fusions by site-directed mutagenesis of the wild-type fusion construct. These mutants had replaced the arginine residue at position 62 of the wild-type SulA with either histidine (PrtA-SulA62H) or serine (PrtA-SulA62S). The function of each fusion construct was tested by two methods to confirm that the wild-type SulA domain retained its biological activity and that the mutant forms were defective. First, the concentration of IPTG which induced expression of lethal levels of each protein was examined. The lowest level of IPTG tested (0.0078 mM) was sufficient to induce significant killing of cells expressing the wild-type PrtA-SulA, while growth of cells expressing the PrtA-SulA62H or PrtA-SulA62S tolerated IPTG at up to 0.25 mM (Fig. 4).

FIG. 4.

Toxic expression of PrtA-SulA fusions. Strain HMS174(DE3) was transformed with 0.5 μg of the indicated plasmid DNA: pET11-PrtASulAwt (wt), pET11-PrtASulA62H (62H), or pET11-PrtASulA62S (62S). The transformed cells were grown for 2 h at 37°C and then diluted in twofold steps in a 96-well plate. From each dilution 10-μl aliquots were spotted onto each of a series of 10 plates containing Luria-Bertani medium plus ampicillin (50 μg/ml) and 0, 0.0078, 0.0156, 0.0234, 0.0312, 0.0468, or 0.0625 mM IPTG. After incubation overnight at 37°C, colonies were counted for each IPTG concentration. The transformation efficiency was determined relative to that achieved with no IPTG and is plotted as the surviving fraction.

Second, the ability to induce filamentation when expressed in cells was examined. Figure 5 shows that cells expressing PrtA-SulA are blocked in division, resulting in filamentous growth, whereas controls which express the mutant SulA fusion proteins are normal rods.

FIG. 5.

Effect of PrtA-SulA alleles on cell division. The cloned SulA alleles are as indicated; “none” refers to the vector lacking SulA. All panels show cells after IPTG-induced expression of the plasmid-encoded gene. Strain HMS174(DE3) was transformed with pET11a, pET11-PrtASulAwt, pET11-PrtASulA62H, and pET11-PrtASulA62S plasmids. One colony from each plate was inoculated into minimal medium (33) and 50 μg of ampicillin per ml at 37°C, then grown to an OD₆₀₀ of 0.2, and then induced with 0.5 mM IPTG for 4 h. Cells from 0.5-ml samples of each well were harvested, washed with 100 mM NaCl in 5 mM Tris-Cl (pH 7.4), and resuspended in 0.5 ml of water with 10 μM Syto-17 red fluorescent dye. Aliquots (10 μl) were spread on polylysine-coated glass slides and fixed with Permount as coverslips were mounted. Slides were examined with a Zeiss Axiovert inverted stage microscope with a ×100 oil immersion objective lens and a rhodamine fluorescence filter set. Exposure was for 4 s on ISO 400 film.

SulA inhibits FtsZ GTPase.

An earlier report by Higashitani et al. (19) failed to detect any effect of a maltose binding protein-SulA fusion protein on FtsZ GTPase activity. Since the PrtA-SulA fusion proteins carry a smaller additional domain (ca. 14 kDa), we considered it important to investigate their effects on the GTPase activity of purified FtsZ protein. Each purified PrtA-SulA protein was preincubated with wild-type FtsZ protein prior to measuring GTPase activity (Fig. 6). Wild-type PrtA-SulA showed a significant level of inhibition of FtsZ GTPase activity. In contrast, neither the PrtA-SulA62H nor the PrtA-SulA62S mutant proteins exhibited an effect on GTPase activity. Preincubation of FtsZ led to a lag in its subsequent GTPase activity.

FIG. 6.

Effects of PrtA-SulA on FtsZ GTPase activity. The indicated PrtA-SulA fusion (5 μM) was each preincubated with 2 μM FtsZ for 5 min before the addition of GTP to the assay. The assays were followed for 20 min, withdrawing and quenching samples for analysis. The GTPase assay was done as described in Materials and Methods.

SulA inhibits FtsZ polymerization.

The effect of the PrtA-SulA proteins on polymerization of FtsZ induced by DEAE-dextran was examined. Wild-type PrtA-SulA was a potent inhibitor of FtsZ polymerization (Fig. 7), whereas the two mutants, PrtA-SulA62H and PrtA-SulA62S, did not inhibit FtsZ polymerization. Since the FtsZ2 mutant was originally isolated as SulA insensitive, the effect of PrtA-SulA on FtsZ2 polymerization was compared with that of the wild type. Again, the results parallel the in vivo observations; although polymerization of the wild-type FtsZ protein was inhibited, there was only minimal effect on the FtsZ2 mutant (Fig. 7).

FIG. 7.

Effects of PrtA-SulA fusion proteins on FtsZ polymerization. Wild-type FtsZ (Z) or FtsZ2 (Z2) was present where indicated at a fixed concentration of 2 μM. Increasing amounts of wild-type PrtA-SulA protein (wt) or indicated mutant SulA fusion proteins (62S and 62H) were preincubated with FtsZ or FtsZ2 for 5 min at 37°C before polymerization was initiated by the addition of DEAE-dextran to 100 μg/ml. The sedimentation assay was then done as described in Materials and Methods. The results have been normalized to facilitate comparison of the data for FtsZ2 (95% sedimented in the absence of any SulA) and wild-type FtsZ (86% sedimented in the absence of any SulA).

Polymerization reactions in the presence of PrtA-SulA fusions were also examined by electron microscopy (data not shown). Wild-type FtsZ formed hundreds of tubules per grid square in the presence of either SulA mutant fusion; these tubules were indistinguishable from the polymers formed in the absence of SulA in number and form. This finding contrasted dramatically with the complete absence of visible polymers in the presence of equimolar wild-type PrtA-SulA, where at least a dozen grid squares were examined in each duplicate grid from four independent experiments. As judged by electron microscopy, FtsZ2 polymerization was not affected even by wild-type SulA. The FtsZ2 structures observed were indistinguishable from those seen in the absence of SulA.

Polymerization was distinctly more sensitive to SulA than was GTPase to SulA (Fig. 8). It required stoichometric levels of SulA (2 μM) to inhibit FtsZ GTPase activity, suggesting that a 1:1 complex may be formed and implying that the affinity may only be in the 1-μM range. In contrast, polymerization of 2 μM FtsZ was inhibited 50% by PrtA-SulA at concentrations as low as 0.3 to 0.5 μM, a finding corresponding to one-quarter or fewer SulA molecules per FtsZ molecule. The affinity estimate from the GTPase studies suggests that the actual number of SulA molecules actively involved in the inhibition of FtsZ polymerization may be considerably lower than one per four FtsZ molecules.

FIG. 8.

Direct comparison of the effects of PrtA-SulA fusion proteins on FtsZ polymerization and on FtsZ GTPase activity. The wild-type PrtA-SulA fusion was titrated into reactions for GTPase or polymerization; in each case reactions were initiated by the addition of FtsZ wild-type protein. The assays were performed as described in Materials and Methods.

The order of addition of SulA to the polymerization reaction was investigated to ascertain whether SulA can cause the disassembly of intact FtsZ polymers. SulA was either preincubated with FtsZ before addition of DEAE-dextran or SulA was added last, after the DEAE-dextran; in either case it was inhibitory. Indeed, when polymers of FtsZ had already formed, the addition of equimolar SulA caused a rapid depolymerization within 5 min of its addition to FtsZ (data not shown). Shorter times could not readily be monitored by the sedimentation assay, so this depolymerization might be much more rapid.

SulA interaction with FtsZ.

Direct binding of the PrtA-SulA fusion to FtsZ was confirmed by affinity column chromatography. FtsZ and PrtA-SulA were mixed in the reaction buffer at 37°C for 15 min and subsequently passed over an IgG-Sepharose column which bound the PrtA-SulA fusion, presumably via the protein A domain. After the column was washed, the PrtA-SulA and any bound FtsZ were eluted and quantified by densitometry of SDS-polyacrylamide gels versus standards. All detectable PrtA-SulA fusion protein was bound by IgG, regardless of the SulA allele. Table 1 shows that the affinity column bearing wild-type SulA retained a significant amount of FtsZ, whereas no FtsZ was detected in the bound fractions when one of the mutant SulA alleles replaced the wild type. The extensive wash step makes this method prone to underestimating the true extent of complex formation. It cannot be concluded that the mutant SulA proteins fail to bind FtsZ but rather that any interaction is much weaker than that of FtsZ with wild-type SulA. It was also seen that FtsZ2 mutant protein bound to SulA to give a yield very similar to that of wild-type FtsZ. This appears somewhat paradoxical in view of the ability of FtsZ2 to polymerize in the presence of SulA, but it is in agreement with a report that these two proteins can interact in the yeast two-hybrid system (20).

TABLE 1.

Affinity copurification of SulA and FtsZ^a

Fusion to PrtA	FtsZ protein	Amt of FtsZ bound (% FtsZ/SulA)	Amt of SulA cosedimented (% SulA/FtsZ)
None	FtsZ	0b	0b
None	FtsZ2	0b	0
SulA	FtsZ	16	0
SulA62S	FtsZ	0	0b
SulA62H	FtsZ	0	0b
SulA	FtsZ2	12	0b

^aIn the affinity column experiments, all detectable PrtA-SulA fusion protein was bound to the column. The fraction of FtsZ bound is expressed as a percent molar ratio relative to the bound PrtA-SulA. In the cosedimentation assays, the amount of PrtA-SulA is expressed as a percent molar ratio relative to the pelleted FtsZ protein. See also Materials and Methods. 

^bThe limits of detection were 1% for FtsZ or 2% for SulA. 

The interaction between the FtsZ2 mutant protein and SulA provided a way to determine whether SulA can bind efficiently to FtsZ subunits within the polymer. We tested whether PrtA-SulA was cosedimented with the FtsZ2 polymer. Such cosedimentation should be more sensitive for detecting interaction than is the affinity copurification procedure since there is no wash step as the polymers are sedimented through a solution containing SulA protein. Despite the sensitivity of the method, cosedimentation of SulA dependent on FtsZ was not detectable with a variety of ratios of SulA either to wild-type FtsZ or to mutant FtsZ2 protein (Table 1). The level of SulA cosedimenting with FtsZ2 polymers provides the most sensitive detection method since polymerization is affected little even by excess SulA. In such FtsZ2 assays, as little as 2% of the input level (2 μM) of SulA could have been measured in the pellet, but none was detected above the background level (1%) of soluble protein trapped within the pellet of FtsZ2 polymer. This suggests that the SulA protein does not bind tightly to polymerized FtsZ2 protein along the entire length of the polymer. The possibility that SulA binds specifically to one or both ends of the FtsZ polymer is not excluded.

DISCUSSION

The results presented here document the inhibition of FtsZ polymerization by SulA, which likely leads directly to the arrest of cell division. Wild-type and mutant forms of these two proteins have been compared in vitro and in vivo, providing evidence in support of this notion.

Fusions of SulA protein were used to circumvent solubility problems encountered with the native protein. The SulA fusion proteins were all tested for biological function in blocking cell division. Only the wild-type SulA fusion was a potent division inhibitor. Inhibition of FtsZ polymerization was observed only for the fusions of the functional wild-type SulA protein and not for the mutant SulA proteins. This correlates with our in vivo results. The inability of the wild-type SulA to inhibit the FtsZ2 protein is consistent with the selection used to isolate the ftsZ2 allele as SulA resistant.

Two extreme possibilities might be considered for the cellular role of FtsZ: (i) GTPase or signaling associated with GTP and GDP binding is the critical aspect of FtsZ function in the cell, or (ii) the ability to polymerize is the most crucial factor for FtsZ function during division. The properties of FtsZ2 support the idea that FtsZ acts as a polymer in the cell. FtsZ2 exhibits a low level of GTPase activity, as low as or lower than the GTPase activity of the SulA-inhibited wild-type FtsZ protein. If GTPase activity were the primary requirement for FtsZ function, then the ftsZ2 mutant should be expected to resemble a loss-of-function mutation. In contrast, FtsZ2 is fully capable of polymerization under our assay conditions even when SulA is present. If polymerization is the crucial property of FtsZ, then the ftsZ2 mutants should be able to divide. The phenotypes of ftsZ2 mutants best fit a role for the polymer.

Although mutants bearing a single copy of the ftsZ2 gene are not viable, it requires only a modest increase in expression of FtsZ2 to grow at permissive temperature (4). If the level of expression were increased even by sevenfold, then the corresponding strains expressing increased wild-type FtsZ protein would be expected to form filaments and minicells (41), whereas this is not reported to be the case (4). Reducing wild-type FtsZ levels by more than two- or threefold results in filamentous growth (14). Thus, despite the 200-fold reduction in the amount of FtsZ2 GTPase, the mutant protein is capable of almost wild-type function in division at 30°C. These observations are not readily compatible with a primary function for FtsZ as a GTPase.

The extremely low level of FtsZ2 GTPase activity is intriguing. The structure of the Methanococcus jannaschii FtsZ protein (25), together with the high amino acid identity, suggests that a simple threading of the E. coli FtsZ amino acid sequence may provide a plausible approximation to the structure of the E. coli FtsZ protein. Such a model would place the FtsZ2 mutation (aspartic acid 212 to glycine) at the opposite end of the FtsZ protein from the GTP-binding site. It is possible that this amino acid change causes an overall alteration in the conformation of FtsZ, thereby affecting the GTPase activity of the protein. However, a second possibility is suggested by a comparison of FtsZ with the structure of the tubulin polymer determined by electron diffraction (31), where the equivalent residue is in close proximity with the nucleotide-binding site of the neighboring tubulin molecule in the protofilament. The similarities between FtsZ and tubulin at the level of molecular structure and also in their protofilament arrangement at lower resolution (12) suggest that similar contacts might occur between FtsZ promoters in a protofilament. Thus, it is possible that the aspartic acid at position 212 may provide or stabilize part of the active site of the neighboring FtsZ subunit. The GTPase sites in the Ras-GAP and Rho-GAP complexes (35, 37) provide a precedent for this type of interaction.

The FtsZ2 protein copurified with SulA as an affinity ligand, with a yield almost as high as when wild-type FtsZ copurified with SulA. This observation provides direct biochemical confirmation of the FtsZ2-SulA interaction detected by the yeast two-hybrid system (20). Despite the affinity copurification of FtsZ2 protein with SulA, we did not detect significant cosedimentation of SulA with FtsZ2 polymers, a less-stringent condition in that it omits any wash step.

There are several possible explanations for this apparent discrepancy. First, SulA may interact preferentially with unpolymerized FtsZ rather than the bulk FtsZ polymer. In this case SulA would shift the equilibrium between the FtsZ polymer and free FtsZ monomer by depleting the level of monomer to form a SulA-FtsZ complex. However, this hypothesis cannot easily explain why substoichiometric levels of SulA were able to inhibit FtsZ polymerization so dramatically.

A second possibility is that SulA binds to an FtsZ subunit at one end of the polymer so as to destabilize that end and thereby favor depolymerization. Such a site might plausibly overlap one of the two contact surfaces of FtsZ involved in polymerization, so that it is only accessible at one end of the polymer but is exposed on each FtsZ monomer. This might explain the sensitivity of FtsZ polymerization to substoichiometric levels of SulA and also the rapid depolymerization of FtsZ induced by addition of SulA. The limits of detection in our assays allow the possibility that SulA interacts with one or both ends of the polymer. The possibility of active depolymerization of FtsZ by SulA was first suggested by in vivo results (3).

Inhibition of FtsZ GTPase requires stoichiometric levels of SulA, suggesting that a 1:1 complex of FtsZ and SulA may be responsible for the reduced level of GTPase. Such a SulA-FtsZ complex is consistent with the binding data. The inhibition of the polymerization assay by substoichiometric levels of SulA may not be due to a different kind of SulA-FtsZ interaction but may reflect the difference in the sensitivity of the measurements. The conditions of the sedimentation assay did not lead to detectable sedimentation of DnaB hexamers (M_r of 300,000 for the hexamer [data not shown]), so that it seems likely that the FtsZ polymers which readily sediment consist of more than six protomers of FtsZ. In such a case, one SulA molecule per 10 or more FtsZ molecules may be sufficient to disrupt detectable polymers in the sedimentation assay. In agreement with this approximate size limit, no significant assemblies of FtsZ were observed by electron microscopy when SulA was present, but where FtsZ minirings of approximately 15 to 20 FtsZ promoters are readily detected by electron microscopy.

This discussion would be incomplete without considering the contrast between our observations of SulA inhibition of FtsZ GTPase activity and those of an earlier report (19) in which no effect of SulA on GTPase was seen. Several explanations might be considered. First, the differences in GTPase reaction conditions may explain the different observations. Second, the larger maltose binding protein fusion partner used in the earlier study may account for the differences; perhaps it allows binding of SulA to FtsZ but hinders a second interaction that mediates the inhibitory effect of SulA. Finally, we noted that the GTPase activity of FtsZ is significantly less sensitive to SulA than is the polymerization of FtsZ. The experiments of Higashitani et al. (19) may have used SulA at a concentration which is too low to cause significant inhibition of FtsZ GTPase.

Earlier studies had identified the ability of DEAE-dextran to promote the polymerization of FtsZ (12, 30). These observations have been extended by the present study to provide a quantitative biochemical assay at a physiological pH. The behavior of wild-type and mutant FtsZ and SulA proteins argues that the DEAE-dextran polymerization assay reflects at least some of the important cellular conditions. Nevertheless, caution is required since this is a different situation from that in the bacterial cell.

While DEAE-dextran represents an artificial condition, we speculate that it may mimic the effects of a factor which normally nucleates or stabilizes FtsZ polymers in the cell. Such a factor is expected to be located in the inner membrane and may be a known division protein, such as the recently identified ZipA protein (17). It is plausible that FtsZ does not polymerize efficiently under physiological conditions without a positively charged factor to nucleate or stabilize the polymeric form. It is noteworthy that the cytoplasmic domains of most Fts proteins carry a significant net positive charge. Although none of the other Fts proteins is essential for localizing FtsZ to a ring, the stability or frequency of the FtsZ rings is reduced in mutants affected in a number of fts genes (1, 24, 32).

The availability of a purified FtsZ polymerization assay which responds to authentic cellular regulators provides a powerful tool for future studies of the molecular mechanisms of division and its control.