Cell Motility and the Cytoskeleton 40:71–86 (1998) FtsZ From Escherichia coli, Azotobacter vinelandii, and Thermotoga maritima—Quantitation, GTP Hydrolysis, and Assembly Chunlin Lu, Jesse Stricker, and Harold P. Erickson* Department of Cell Biology, Duke University Medical Center, Durham, North Carolina We have cloned the ftsZ genes from Thermotoga maritima and Azotobacter vinelandii and expressed the proteins (TmFtsZ and AzFtsZ) in Escherichia coli. We compared these proteins to E. coli FtsZ (EcFtsZ), and found that several remarkable features of their GTPase activities were similar for all three species, implying that these characteristics may be universal among FtsZs. Using a calibrated protein assay, we found that all three FtsZs bound 1 mole guanine nucleotide per mole FtsZ and hydrolyzed GTP at high rates (.2 GTP per FtsZ per min). All three required magnesium and a monovalent cation for GTP hydrolysis. Previous reports showed that EcFtsZ (and some other species) required potassium. We confirmed this specificity for EcFtsZ but found that potassium and sodium both worked for Az- and TmFtsZ. Specific GTPase activity had a striking dependence on FtsZ concentration: activity (per FtsZ molecule) was absent or low below 50 µg/ml, rose steeply from 50 to 300 µg/ml and plateaued at a constant high value above 300 µg/ml. This finding suggests that the active state requires a polymer that is assembled cooperatively at 50–300 µg/ml. A good candidate for the active polymer was visualized by negative stain electron microscopy—straight protofilaments and protofilament pairs were seen under all conditions with active GTPase. We suggest that the GTP hydrolysis of FtsZ may be coupled to assembly, as it is for tubulin, with hydrolysis occurring shortly after an FtsZ monomer associates onto a protofilament end. As a part of this study, we determined the concentration of EcFtsZ and TmFtsZ by quantitative amino acid analysis and used this to standardize the bicinchonic acid colorimetric assay. This is the first accurate determination of FtsZ concentration. Using this standard and quantitative Western blotting, we determined that the average E. coli cell has 15,000 molecules of FtsZ, at a concentration of 400 µg/ml. This is just above the plateau for full GTPase activity in vitro. Cell Motil. Cytoskeleton 40:71-86, 1998. r 1998 Wiley-Liss, Inc. Key words: tubulin; dynamic instability; nucleation; cell division; bicinchonic acid; quantitative Western blotting INTRODUCTION The bacterial cell division protein FtsZ has been found in more than three dozen prokaryotic species, including archaebacteria, and in chloroplasts [Erickson, 1997]. It is a major cytoskeletal protein of bacteria, forming a ring that constricts at the site of septation [Addinall et al., 1996; Ma et al., 1996]. FtsZ is a homologue of tubulin, as demonstrated first by sequence r 1998 Wiley-Liss, Inc. Contract grant sponsor: National Institutes of Health; Contract grant number: GM28553 *Correspondence to: Harold P. Erickson, Department of Cell Biology, Duke University Medical Center, Durham, NC 27710; E-mail: H.Erickson@cellbio.duke.edu Received 26 November 1997; accepted 21 January 1998 72 Lu et al. alignment [Mukherjee and Lutkenhaus, 1994], and recently confirmed by the close similarity of the atomic structures [Löwe and Amos, 1998; Nogales et al., 1998]. The homology is functional as well, as Escherichia coli FtsZ (EcFtsZ) can assemble in vitro into protofilament sheets with a lattice similar to that of the microtubule wall [Erickson et al., 1996]. FtsZ protofilaments can also adopt a sharply curved conformation, forming minirings very similar to tubulin rings [Erickson et al., 1996; Erickson and Stoffler, 1996]. The structural homology of FtsZ and tubulin polymers suggests that the cytoskeletal function of these proteins has been conserved from bacterial FtsZ to eukaryotic tubulins [Erickson et al., 1996; Erickson, 1997]. Both tubulin and FtsZ bind and hydrolyze GTP. The GTPase activity of tubulin is tightly coupled to assembly. In the absence of assembly the rate of hydrolysis is very low, but after a subunit enters the microtubule lattice its bound GTP is rapidly hydrolyzed [O’Brien et al., 1987]. EcFtsZ purified from overexpression in bacteria has demonstrated a very high level of GTPase [RayChaudhuri and Park, 1992; de Boer et al., 1992; Mukherjee et al., 1993]. In the present study we have reinvestigated and extended the characterization of the GTPase of EcFtsZ. As a part of this study, we have calibrated the assay for FtsZ protein concentration and determined the concentration in E. coli. In addition, we have cloned the ftsZ genes from Azotobacter vinelandii and Thermotoga maritima, expressed these proteins in E. coli (AzFtsZ and TmFtsZ), and compared the GTPase and polymer state of the FtsZ from the three species. METHODS Cloning AzFtsZ and TmFtsZ Two degenerate oligonucleotide primers were designed based on the most highly conserved regions, NTDNQA and GGGTGTG of the N-terminal portion of FtsZ (forward primer, 58-AA(A/T)AC(G/A/T/C)GA(C/T)GC(G/A/T/C)CA(A/G)GC-38 reverse primer, 58-GT(G/A/ T/C)CC(G/A/T/C)GT(G/A/T/C)CC(G/A/T/C)CC(G/A/T/ C)CC-38). Genomic DNA from Thermotoga maritima was generously provided by Dr. Michael Adams (University of Georgia, Athens, GA). Polymerase chain reaction (PCR) products were separated on an agarose gel, and fragments in the range of 200–500 base pairs (bp) were isolated and cloned. A 350-bp fragment was identified that showed significant homology to ftsZ of other species. Thermotoga genomic DNA was cut with EcoRI and circularized, and inverse PCR was performed as described by Ochman and Hartl  with erratic but eventually successful results. Each attempt at inverse PCR gave many clones with unrelated sequence, but some showed homology to ftsZ of other species, extending the sequence in the 58 and 38 directions. New primers were designed near the limits of these extended sequences, and the inverse PCR was repeated. Sequences were eventually found that extended past the initiation and stop codons, as judged by homology with ftsZ sequences of other species. The final sequence was obtained from the pET-FtsZ plasmid (described below), which was produced with a single PCR reaction from genomic DNA using the error-correcting Pfu polymerase. The same PCR primers were used to amplify genomic DNA of Azotobacter vinelandii (strain CA, provided by Dr. Paul E. Bishop, North Carolina State University, Raleigh). A 350-bp PCR product was then used to screen a cosmid library, provided by Dr. Limin Zhang (Virginia Polytechnic Institute, Blacksburg, VA). A positive cosmid was identified and digested by EcoRI. A ,7-kbp fragment that hybridized with the PCR probe was gel-purified and subcloned into the SuperCos 1 cosmid vector (Stratagene, LaJolla, CA). The ftsZ gene was sequenced in both directions starting from the region identified by PCR, and the full-length ftsZ was obtained as judged by comparison to the ftsZ sequences of other species. Construction of Recombinant Expression Vector for TmFtsZ and AzFtsZ Once we had identified the initiation and stop sequences of Thermotoga ftsZ, we designed primers to amplify the full-length coding sequence, adding an NdeI site at the 58 end and a BamHI sequence at the 38 end. PCR amplification used 2.5 units of Pfu DNA polymerase (Stratagene, LaJolla, CA) in a 50-µl reaction mixture containing 0.2 mM dNTP, 1 µM primers, 0.5 µg Thermotoga genomic DNA and the buffer recommended by Stratagene. The predicted 1,070-bp fragment was isolated from an agarose gel, digested with NdeI and BamHI at 37°C for 24 h, gel purified again, and then ligated into NdeI and BamHI sites of expression vector pET-11b (Novagen, Madison, WI). Both strands of the resulting recombinant vector pET-TmFtsZ were sequenced from the synthetic NdeI site through the BamHI site to determine the final version of the Thermotoga ftsZ sequence. The same procedure was used to produce a pET-11 expression vector for AzFtsZ. The pET expression plasmid for EcFtsZ was provided by Dr. David Bramhill, Merck [Bramhill and Thompson, 1994] and was also duplicated in our laboratory. Expression and Purification of Ec-, Az-, and TmFtsZ EcFtsZ, AzFtsZ, and TmFtsZ were successfully overproduced in E. coli strain BL21(DE3) transformed with pET-11 expression vectors containing the coding region of the three genes. Cultures for all three were GTPase and Assembly of FtsZ grown in LB medium at 37°C to A600 ,1.2 and induced by adding 0.25–0.5 mM IPTG. We also induced some cultures at A600 ,0.7 and found a slightly lower yield of FtsZ. The protein appeared identical at the two levels of expression; in particular the ratio of protein precipitated at 20% and 25% ammonium sulfate (see below). After a 3-h induction, bacteria from a 1-L culture were centrifuged and suspended in 20 ml resuspension buffer (0.05 M Tris, 0.1 M NaCl, 1 mM EDTA, 1 mM PMSF, pH 8.0). Lysozyme was added to 0.4 mg/ml, and the mix was incubated on ice for 2 h; MgCl2 was added to 5 mM and the sample was frozen overnight at 220°C. The bacteria were thawed and sonicated twice for 30 s on ice. DNase I was added to 10 µg/ml and incubated 60 min at 4°C. Cell walls and insoluble debris was removed by centrifugation. All three FtsZ proteins were soluble in E. coli cytoplasm, and represented more than one-half of the total soluble protein. In earlier preparations of FtsZ, we added Triton X-100 to 1% after 30 min with lysozyme, continued agitation for 5 min and froze the mixture. However, the triton caused the ammonium sulfate precipitate to partially float when centrifuged, instead of forming a compact pellet. This was a problem especially for TmFtsZ whose precipitation required higher ammonium sulfate. We have subsequently eliminated Triton from the lysis buffer and have found that freezing and sonication efficiently lyses the bacteria. After centrifuging the lysed bacteria, EcFtsZ was precipitated from the bacterial supernatant with ammonium sulfate. Initially we used 20% saturated ammonium sulfate, which precipitated virtually all the FtsZ and almost no bacterial proteins in the presence of 1% Triton. After eliminating Triton we found that less than one-half of the EcFtsZ was precipitated at 20% saturated ammonium sulfate, and it had a substantially lower GTPase activity than protein precipitated at 25%. We then modified our purification protocol. We first did a cut at 20% saturated ammonium sulfate, discarding the pellet (about 30–40% of the FtsZ). We then precipitated the remaining EcFtsZ by raising the ammonium sulfate to 25% saturation. The pellet was resuspended in one-half the original volume of resuspension buffer supplemented with 10 mM Mg and re-precipitated with 25% saturated ammonium sulfate. At this stage, EcFtsZ appeared pure on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDSPAGE), and we found that ion-exchange chromatography on DEAE Sephacel or mono-Q did not result in any apparent increase in purity. For the experiments reported in this paper, EcFtsZ was prepared by the 20–25% ammonium sulfate precipitation, with no additional steps. TmFtsZ required 38% saturated ammonium sulfate for precipitation, and many host proteins were pelleted 73 with it. To separate contaminating proteins from TmFtsZ, a heating step was introduced to denature and precipitate the E. coli proteins. We found that TmFtsZ was substantially stabilized when 1 mM GTP and 5 mM Mg were added before heating. With GTP and Mg, TmFtsZ remained soluble after 20 min at 70°C or 10 min at 80°C, while most contaminating proteins precipitated. For purification, a 2- to 3-ml sample of TmFtsZ in a centrifuge tube was heated to 80°C for 2 min, cooled on ice for 20 min, and then centrifuged. As a final purification step, the TmFtsZ was chromatographed on a mono-Q column, where it eluted at 0.37 M NaCl. AzFtsZ was purified by two cycles of precipitation at 20% saturated ammonium sulfate. Eventually, all FtsZ proteins were resuspended in the resuspension buffer with 10 mM Mg, aliquoted, and stored at 280°C. Calibration of Protein Concentration Quantitative amino acid analysis of bovine serum albumin (BSA), EcFtsZ, and TmFtsZ was performed by Dr. Ida Thorgersen (Macromolecular Structure Facility, Duke University Medical Center). BSA concentration was also checked by ultraviolet (UV) spectroscopy. These calibrated standards were used to calibrate the bicinchonic acid (BCA) colorimetric assay (Pierce Biochemicals, Rockford, IL) [Smith et al., 1986]. Antibody Production Rabbits were inoculated with 200 µg of GST-FtsZ fusion protein in 2 ml saline mixed with 2 ml complete Freund’s adjuvant. One rabbit (FtsZd-5434) was immunized with a protein band that was cut out from an SDS gel (,2 ml gel was equilibrated in saline and emulsified with Freund’s adjuvant). Another rabbit (FtsZn-5435) was injected with soluble, native protein that had been purified by ammonium sulfate precipitation and affinity chromatography on a glutathione column. Four boosts of 50 µg protein in 2 ml saline plus 2 ml incomplete Freund’s adjuvant were given over 10 weeks. All bleeds for both antisera stained a single band, of varying intensities, at ,40 kDa, in Western blots of whole E. coli. Bleed 4 of the FtsZn-5435 antiserum was further purified by affinity binding to EcFtsZ coupled to CNBr-Sepharose. A total of 1.44 mg of antibody was purified from 5 ml antiserum. Quantitative Western Blotting Escherichia coli BL21 cells were used for determination of normal FtsZ levels. Cells were grown in LB at 37°C, shaken at 250 rpm, to an OD600 of 0.5–0.7. Cells were then killed with 0.02% sodium azide and put on ice for 30 min. They were then counted with a hemocytometer under 4003 magnification using phase contrast optics. Several cultures were also diluted and plated on LB-agar plates before fixing so as to carry out viable 74 Lu et al. counts. In each case, viable counts agreed with hemocytometer counts within 10%. Samples containing a known number of cells were lysed in SDS buffer and separated on 12% SDSpolyacrylamide reducing gels, transferred to Immobilon membrane (Millipore, Bedford, MA), and blocked with 5% skim milk/TTBS (100 mM Tris pH 7.4, 150 mM NaCl, 0.1% Tween) overnight. Primary incubation was with 0.23 µg/ml FtsZn-5435 affinity purified polyclonal antibody in 0.25% gelatin/TTBS for 120min. Secondary incubation was with 2 µCi/blot 125I-labeled protein A (ICN, Costa Mesa, CA) in 0.25% gelatin/TTBS for 45min. Blots were washed and placed on phosphoimager screens (Hypercassette, Amersham, Arlington Heights, IL) for periods ranging from 3 h to 16 h, then scanned and digitized with a phosphoimager (Fujix Bas1000). The intensity of lanes on these blots was determined with the software program MacBas 1.0 (Fuji). For quantitative Western blotting, blots were run in sets to ensure consistent results. Three identical blots containing a set of standards and a set of unknowns and a blot containing two sets of standards were run concurrently. The dual standards blot allowed confirmation of consistency within blots, showing that all areas on the blot were equally sensitive. The multiple blots containing unknowns allowed confirmation of consistency between blots. Any set of blots where identical lanes on the dual standards blot had intensities differing by more than 10% was discarded. Similarly, in the case in which the unknowns blots gave intensities differing by more than 10%, they would be discarded. If two of the three unknowns blots agreed with each other, only the third was discarded. If all three blots disagreed, all three were discarded. GTP Hydrolysis FtsZ was desalted, and its buffer was exchanged using prepacked Econo-Pac 10DG columns (Bio-Rad, Hercules, CA) in 50 mM Tris-HCl, pH 7.5. Three reaction buffers were used. Most assays were done in TKM7.5 (50 mM Tris-HCl, pH 7.5, 50 mM KCl, 5 mM Mg acetate). The second buffer was PB7.7 (physiological buffer: 35 mM MOPS, pH 7.7, 80 mM Glutamate, 350 mM potassium acetate, 2.5 mM Mg acetate, 50 mM trehalose, 2 mM putrescine), which mimics the ionic composition of the E. coli cytoplasm [Cayley et al., 1991]. We found that glutamate, trehalose and putrescine had no effect on GTPase or assembly, so this buffer is very similar to TKM7.5, except for the higher potassium concentration. MEMK6.5 (100 mM MES, pH 6.5, adjusted with KOH, 1 mM EGTA, 5 mM Mg acetate) was a lower pH buffer used for some assembly and GTPase assays. Buffers normally contained 0.5 mM GTP (except as noted), and 0.5 µCi (a-32P) GTP (3,000 Ci/mmol). GTP from Amer- sham was uniformly of high quality when analyzed by thin layer chromatography, and was used for all assays; several lots obtained from ICN contained only GDP. Additives or alterations in the buffer are indicated in the text. Reaction buffer for TmFtsZ was prewarmed to the reaction temperature and the reaction was initiated by adding prewarmed FtsZ to give the indicated concentration. At indicated intervals, 1 µl was withdrawn and applied to polyethyleneimine cellulose thin layer chromatography plates (10 3 10 cm, Polygramt CEL 300 PEI/UV254, Macherey-Nagel, Germany). The spots dried within 1–2 min, which we assumed arrested the GTPase. Plates were developed in 0.75 M KH2PO4, pH 3.4, dried, and exposed to a phosphoimager plate (IP, FUJIX) for 10–30 min. The imager plate was screened and data were analyzed using MacBAS1000 v 1.01 software (FUJIX, BAS 1000Mac, Fuji Photo Film). To avoid errors due to evaporation of sample during the reaction or imprecision in the amount of sample loaded, the data were always read as the ratio of GDP/GTP; this ratio, multiplied by the known amount of GTP in the starting solution, gave the amount of GDP produced, which was eventually expressed as moles GDP produced per mole FtsZ. Because we do not understand the mechanism for the GTPase reaction, and because variable results have been reported by different labs, we need to specify the conditions of our assay precisely. The FtsZ was transferred by gel filtration to 50 mM Tris-HCl, pH 7.5 (containing no GTP, Mg, or K), and frozen at a concentration of 1–3 mg/ml. FtsZ was thawed and kept on ice before the experiment. A reaction mixture was prepared containing Tris, KCl, Mg, and GTP to give the desired final concentrations when mixed with FtsZ in a 20-µl reaction mixture. FtsZ was added to the reaction mixture on ice, and the Eppendorf tube was transferred immediately to a 37°C water bath (for Ec- and AzFtsZ). For TmFtsZ, the concentrated FtsZ and the reaction buffer containing GTP were separately prewarmed to 70°C and mixed at the initial time point. Attempts to prewarm EcFtsZ gave erratic results, perhaps because of denaturation of the protein at 37°C in the absence of GTP. Assembly of FtsZ Three buffers were used for assembly studies: TKM7.5, PB7.7, and MEMK6.5 . FtsZ was transferred to assembly buffer using 10-ml prepacked Econo-Pac 10DG columns (Bio-Rad) before assembly. Assembly of TmFtsZ in MEMK6.5 was started by adding GTP to 2 mM and incubating at 70°C. At 5 min after incubation at 70°C, glutaraldehyde (ultrapure TEM grade, Tousimis Research Corporation, Rockville, MD) was added to 1% to fix the polymers. The reaction was kept at 70°C for another 2 min, then cooled to room temperature, and negatively stained specimens were prepared. EcFtsZ (1 mg/ml) was assembled GTPase and Assembly of FtsZ in both MEMK6.5 and PB7.7, while AzFtsZ was polymerized in PB7.7. These reactions were started by adding DEAE dextran (approximate MW 2,000,000, D-5876, Sigma, St. Louis, MO) to 0.6 mg/ml and GTP to 2 mM, and incubated on ice for 10 min, followed by incubation at 37°C for 5 min. EcFtsZ polymers in the absence of DEAE dextran were visualized by negative staining at 0°C and 37°C. Electron Microscopy Before rotary shadowing, FtsZ was sedimented through a 5-ml linear gradient of 15–40% glycerol in 0.1 mM GTP, 0.2 M ammonium bicarbonate. Peak fractions identified by SDS-PAGE were diluted in 40% glycerol, 0.2 M ammonium bicarbonate, sprayed onto freshly cleaved mica, and rotary shadowed with platinum [Fowler and Erickson, 1979]. Assembly of FtsZ in vitro was analyzed primarily by negative stain electron microscopy. Grids coated with a thin carbon film were subjected to glow discharge to render the surface hydrophilic and wettable. A total of 10 µl of the assembled FtsZ was applied to the grid, removed, and drained with filter paper, and 3 drops of uranyl acetate (a 2% aqueous solution, unadjusted pH, filtered through 0.2 µm) were washed over the grid while it was held at a 45° angle. The grid was finally drained with filter paper and stored. Grids were examined with a Philips 301 electron microscope and photographed at 50,0003. RESULTS Sequences of AzFtsZ and TmFtsZ and Expression in E. coli The DNA and deduced amino acid sequences of AzFtsZ and TmFtsZ have been submitted to the GenBank database, accession numbers U65939 and U65944. The amino acid sequence of AzFtsZ is 85% and 83% identical to FtsZ from the closely related Pseudomonas aeruginosa and Pseudomonas putida, and 57% identical to EcFtsZ. TmFtsZ is 45% identical to FtsZ from the gram-positive bacteria Bacillus subtilis, Staphylococcus aureus, and Anabaena sp. It is about 40–42% identical to FtsZ from most gram-negative bacteria (41% to EcFtsZ) and 33– 41% identical to FtsZ from archaebacteria. On a phylogenetic tree the TmFtsZ sequence branches close to the separation of gram-positive, gram-negative, and archaebacteria [Erickson, 1997]. The 351-amino acid sequence of TmFtsZ is one of the shortest of 35 known complete sequences; the only shorter are FtsZ from Haloferax volcanii [Wang and Lutkenhaus, 1996], and the second, probably nonessential FtsZ of Rhizobium meliloti [Margolin and Long, 1994]. All three FtsZ proteins were abundantly expressed as soluble proteins in E. coli. Overexpression of TmFtsZ 75 did not cause filamentation or otherwise alter cell growth or viability either before or after induction by IPTG, suggesting that this protein is inactive in E. coli at 37°C. However, overexpression of EcFtsZ or AzFtsZ caused filamentation. Even without induction by IPTG, primary transformants carrying the EcFtsZ or AzFtsZ recombinant expression plasmid frequently produced smooth filamentous bacteria, apparently due to the small amount of expression from the pET vector. Cell filamentation disappeared after several passages for EcFtsZ-transformed cells, consistent with previous observations [Dai and Lutkenhaus, 1992]. However, the AzFtsZ-transformed cells remained filamentous, suggesting that even a small amount of AzFtsZ inhibits the E. coli division system. This toxicity has been observed for expression of almost all foreign FtsZ in E. coli [Beall et al., 1988; Margolin et al., 1991]. Two Fractions of EcFtsZ With Different GTPase Activities In our earlier experiments, EcFtsZ was purified by two cycles of precipitation with 20% saturated ammonium sulfate. However, further experiments suggested that this EcFtsZ was a mixture of inactive protein and a fully active form. This preparation represented only 30–40% of the total expressed EcFtsZ; it bound only 0.25 moles of guanine nucleotide per mole of EcFtsZ; and it had a lower GTPase activity than any other FtsZ preparation. By contrast, the fraction of EcFtsZ precipitated by 25% ammonium sulfate, following a 20% ammonium sulfate cut, bound 1 mole guanine nucleotide per mole FtsZ and had a much higher rate of GTP hydrolysis. The 20% cut hydrolyzed 0.31 moles of GTP per mole of protein per min, compared with 2.4 moles/min for the 25% cut under the same conditions (see below). We assume that the 25% fraction contained fully active EcFtsZ and used this for all subsequent experiments. To check for possible covalent differences in the two protein preparations, EcFtsZ fractions precipitated at 20% and 25% ammonium sulfate were analyzed by electrospray mass spectrometry. The measured mass was 40,326.9 Da and 40,327.4 Da, for the two preparations, identical to the mass of 40327.4 calculated from the amino acid sequence by the DNAstar (Madison, WI) program. We conclude that neither preparation of EcFtsZ has any post-translational modification. Both fractions also gave identical results upon gradient sedimentation and electron microscopy (see Fig. 5 for micrographs of the 25% fraction), suggesting no difference in polymerization state. Before dismissing the 20% cut as completely uninteresting, we should point out that this sample would assemble into protofilament sheets in MEMK6.5 and PB7.7 buffer; remarkably, this assembly occurred without 76 Lu et al. addition of DEAE dextran. A maximum of 25–30% of the protein in the 20% ammonium sulfate fraction could be pelleted after assembly, compared to 95% of the protein in the 25% ammonium sulfate fraction, but assembly of this later fraction was completely dependent on added DEAE dextran (see below). It is possible that something in the 20% cut is substituting for DEAE dextran. Concentration of FtsZ In Vitro and In Vivo Determining the concentration of FtsZ is not straightforward. UV absorption is usually the simplest and most accurate method [Brennan and Hardeman, 1993; Gill and Von Hippel, 1989], but FtsZ has no tryptophan and almost no tyrosine, and therefore has negligible absorption at 278 nm. Colorimetric assays can give widely different color intensity for different proteins and need to be standardized. We therefore used quantitative amino acid analysis to determine the concentration of samples of EcFtsZ, TmFtsZ, and BSA. (As an incidental result, we independently determined the concentration of BSA by UV absorption and confirmed that the classic extinction coefficient, 0.667 [Foster and Sterman, 1956], is correct, rather than the 7% lower value calculated from the amino acid sequence [Brennan and Hardeman, 1993; Gill and Von Hippel, 1989].) We then compared the three proteins in the BCA assay and found that the ratio of color was 0.75 for EcFtsZ/BSA, and 0.67 for TmFtsZ/BSA. Therefore a substantial correction factor of 1.33 or 1.5 needs to be applied when these proteins are measured in the BCA assay, using BSA as a secondary standard. We also calibrated the Bradford assay, in which the ratio of color was 0.82 for EcFtsZ/BSA, and 0.70 for TmFtsZ/BSA. In our experience, the BCA assay gives much more consistent results. Two previous studies have reported the concentration of FtsZ in E. coli to be 20,000 or 5,000 molecules per cell [Dai and Lutkenhaus, 1992; Pla et al., 1991], but the methodologies were not well described. We therefore determined the concentration of FtsZ in log phase BL21 E. coli by quantitative Western blotting, using the calibrated FtsZ sample described above. We averaged 12 quantitative blots over two separate experiments and two different concentrations of cell lysate. The average value was 14,800 molecules of FtsZ per cell, with a standard deviation of 1,450. We conclude that there are approximately 15,000 molecules of FtsZ in the average log phase E. coli cell, plus or minus 10%. To compare our in vitro assays with conditions in the cell, it is important to know the concentration of FtsZ in the bacterial cytoplasm. The average E. coli cell (strain B/r A, doubling time of 25min) has a length of 3.6 µm [Donachie et al., 1976] and a diameter of approximately 0.94 µm [Trueba and Woldringh, 1980]. These dimensions correspond well with our light microscopy observations (data not shown). Assuming that the bacterial cell is a cylinder capped by two hemispheres, the average cell volume is calculated to be 2.28 µm3. Kubitschek et al  gives the volume of a newborn cell under similar conditions as 1.566 µm3. An average cell under log phase growth would have a volume of (1.566 µm3)/ln2, or 2.26 µm3, which corresponds to our calculated volume. If there are 15,000 molecules of FtsZ in this volume, the intracellular concentration of FtsZ is 10.9 µM, or 440 µg/ml. We noted that the pET expression system for EcFtsZ gives one of the highest expression levels of any protein we have seen, so we determined the concentration of the expressed protein. After 120 min induction we found 3.1 3 106 molecules of FtsZ per cell, and noted also that the cells were 2.4 times as long as normal log phase cells. This gives an intracellular concentration of 36 mg/ml, or ,80 times the average normal concentration of FtsZ. Stoichiometric GTP Binding and Temperature-Dependent Hydrolysis We used our calibrated protein concentration to obtain an accurate measurement of the stoichiometry of GTP binding to the three FtsZs. Starting with a sample of FtsZ in 1 mM GTP, we removed free nucleotide by gel filtration, determined the protein concentration as described, and precipitated the protein with 4.2% perchloric acid. We then determined the concentration of guanine nucleotide by UV spectrophotometry. We obtained a value of 0.99 moles of guanine nucleotide per mol of EcFtsZ. We did not determine whether the nucleotide was GTP or GDP. The measured stoichiometry near 1:1 demonstrates that the affinity of guanine nucleotide is sufficiently high to remain bound during gel filtration. Similar measurements showed 1.03 mol guanine nucleotide per mol AzFtsZ and 1.13 mole guanine nucleotide per mol TmFtsZ. This is the first study to use calibrated protein concentration to demonstrate a 1:1 stoichiometry of guanine nucleotide binding to FtsZ and confirms that the FtsZ proteins purified by our protocols are ,100% active in nucleotide binding. We then assayed the GTPase activity in different buffer conditions. We found that the GTPase activity of both TmFtsZ and AzFtsZ required Mg21 (data not shown), as shown previously for EcFtsZ [RayChaudhuri and Park, 1992; de Boer et al., 1992; Mukherjee et al., 1993] and B. subtilis FtsZ [Wang and Lutkenhaus, 1993]. We measured the GTPase activity of EcFtsZ in two assembly buffers, MEMK6.5 and PB7.7. EcFtsZ showed measurable hydrolysis at 0°C, and a substantial increase at 37°C (Fig. 1a). The activity in PB7.7 was higher than that in MEMK6.5, probably because of the higher potassium concentration, which enhanced activity significantly (see below for details). The GTPase of TmFtsZ showed a GTPase and Assembly of FtsZ 77 FtsZ, or (2) GTP turnover, in moles GDP produced per mole FtsZ per min. Requirement for Sodium or Potassium for GTPase Activity Mukherjee et al.  reported that 100 mM potassium was required for the GTPase activity of EcFtsZ, and 2.0 M KCl was required for GTPase of the halophilic archaebacterium Haloferax volcanii [Wang and Lutkenhaus, 1996]. We investigated how the GTPase of EcFtsZ varied with the concentration of KCl (Fig. 2a). Minimal activation was obtained at 25 mM KCl, and maximal stimulation required 100 mM KCl. Increasing the KCl to 200–400 mM gave no additional enhancement. 400 mM KCl actually produced a several min lag before hydrolysis began. 50 mM KCl gave about half-maximal stimulation of GTPase, and this was used for most subsequent experiments. The enhancement was specific for potassium, since the GTPase in 50 mM NaCl was the same as in Tris alone (Fig. 2b). This finding is consistent with the previous study of EcFtsZ [Mukherjee et al., 1993]. However, we found that FtsZ from Azotobacter and Thermotoga do not have a specific requirement for potassium. As shown in Figure 2c,d, a monovalent cation is required, but sodium is almost as effective as potassium for AzFtsZ and is slightly more effective for TmFtsZ. LiCl was also able to stimulate TmFtsZ GTPase activity (data not shown). GTPase Activity Depends on Concentration of Protein and GTP Fig. 1. GTPase activity of EcFtsZ and TmFtsZ at different temperatures. a: EcFtsZ (1.9 mg/ml) was tested at 0°C for 15 min; the temperature was then shifted to 37°C. GTPase was tested in two buffers. b: TmFtsZ was 0.8 mg/ml in TKM7.5. The plateau is reached when all the GTP is hydrolyzed. Specific activity was expressed as mols GTP hydrolyzed per mol FtsZ at each time point. strong dependence on temperature (Fig. 1b). There was very little activity at 37°C and a modest activity at 50°C. At both 37°C and 50°C, there was a small lag in GTPase, indicated by the acceleration of activity at later times. GTPase was markedly increased and the lag was not detectable at 65°C or higher. Throughout this paper, we express GTP hydrolysis as (1) specific activity, in moles GDP produced per mole Our initial studies showed that the GTPase activity was constant over a 5-fold concentration range, from 200 µg/ml to .1 mg/ml, for each of the three FtsZs. However, previous studies have found that the GTPase activity of FtsZ depends on the protein concentration, implying that self-association may play a role in the hydrolysis reaction [de Boer et al., 1992; Mukherjee et al., 1993; Wang and Lutkenhaus, 1993]. In most cases, the major effect of concentration was a steep loss of GTPase activity at FtsZ concentrations of ,200 µg/ml, so we investigated this range in detail. Time courses of GTP hydrolysis at different FtsZ concentrations are shown in Figure 3a, and the turnover of GTP per FtsZ molecule per min in Figure 3b. Turnover was independent of protein concentration above 300 µg/ml for Ec-, Az-, and TmFtsZ, and for all three proteins the GTP turnover dropped below 200 µg/ml. This is similar to BsFtsZ, which showed a similar plateau above 200 µg/ml, and a steep drop to zero activity at 100 µg/ml [Wang and Lutkenhaus, 1993]. There were two different behaviors at lower concentrations. For Ec- and AzFtsZ there was a lag of 5–20 min before GTP hydrolysis began, confirming two previous reports for EcFtsZ [de Boer et 78 Lu et al. Fig. 2. Effect of monovalent cations on the GTPase activity of Ec-, Az-, and TmFtsZ. Buffers all contained 70 mM Tris-HCl, pH 7.5, 5 mM magnesium acetate, 0.5 mM GTP, and either no additional salt, or KCl or NaCl as indicated. a: Reaction containing 300 µg/ml EcFtsZ and increasing amounts of KCl as indicated. b–d: The reactions contained 50 mM KCl or NaCl as indicated (Tris indicates no additional monovalent cation), plus (b) 650 µg/ml EcFtsZ; (c) 380 µg/ml AzFtsZ; (d) 800 µg/ml TmFtsZ. EcFtsZ and AzFtsZ were at assayed 37°C, while TmFtsZ was at 65°C. The y-axis plots mols GTP hydrolyzed per mol FtsZ at each time point. Each reaction reached a plateau when all the GTP had been hydrolyzed. GTPase activity is indicated by the initial slope, not the height of the curve at plateau. Note that the time scale varies in the different panels. al., 1992; Mukherjee et al., 1993]. For TmFtsZ there was no lag, similar to reports for BsFtsZ [Wang and Lutkenhaus, 1993] and H. volcanii FtsZ [Wang and Lutkenhaus, 1996]. The lag suggests a nucleation phenomenon to generate the active polymers, but it is curious that it is observed in FtsZ from some species but not from others. Our observation in a single experiment of a lag in two species, and not in another, supports the interpretation that these are real species differences and not due to different protocols in different laboratories. We also extend to three new species the observation that GTP turnover per FtsZ is constant above 300 µg/ml, and drops at lower concentrations, suggesting that this is a consistent feature of FtsZ from all bacteria. The experiments depicted in Figure 3 were done at GTP concentrations of 0.15–0.5 mM GTP, the lower concentrations being used for increased accuracy at the lower protein concentrations. Several points were done in duplicate at both 0.15 and 0.5 mM GTP, and there was no significant difference in the GTPase. Figure 4 confirms GTPase and Assembly of FtsZ that there is virtually no difference between 0.15 and 0.5 mM GTP. However, we found surprisingly large increases in GTPase when the GTP concentration was increased to 1–5 mM (Fig. 4). This experiment used 0.3 mg/ml EcFtsZ, which was just above the plateau of the curves in Figure 3, and either 50 or 350 mM KCl. There were two effects of increasing GTP, and these were similar at both KCl concentrations. First, there was no lag at 0.2–0.5 mM GTP, but a 5- to 10-min lag appeared at 1–5 mM GTP. This is particularly evident in the reduced hydrolysis at 5 min at the higher GTP concentrations. Second, the rate of hydrolysis following the lag increased markedly at higher GTP concentration. At each GTP concentration the GTPase was about 2–3 times higher at 350 than at 50 mM KCl, similar to the results in Figure 2. The maximum GTP hydrolysis obtained in our experiments was 26 mols GTP hydrolyzed per min per mol EcFtsZ (at 350 mM KCl and 5 mM GTP). Figure 2 suggests that further increase in KCl would have no effect, but Figure 4 suggests that hydrolysis rate would be increased still further at higher GTP. Note that the GTPase seems to be directly proportional to GTP over the concentration range 0.5–5 mM, but the curves do not extrapolate to zero. At both 50 and 350 mM KCl, the GTP turnover extrapolates to 2 min21 at low GTP concentration. We considered the possibility that the enhanced GTPase at 1–5 mM GTP could be a nonspecific effect of nucleoside triphosphate. We tested the GTPase activity of EcFtsZ in 0.5 mM GTP, with and without 4.5 mM ATP. The GTPase activity was not affected by the 4.5 mM ATP. We tested whether GDP, the product of GTPase, could inhibit the GTPase activity of the FtsZ proteins. In 0.15 mM GTP and 300–500 µg/ml EcFtsZ, the addition of an equal amount of GDP inhibited the GTPase activity somewhat, and addition of a 5-fold excess of GDP inhibited to 42% of full activity. AzFtsZ was more sensitive to GDP than the others, and showed 68% (with 0.15 mM GDP) and 20% (with 0.75 mM GDP) of full activity. TmFtsZ gave 78% activity in 5-fold excess GDP. All Three FtsZs Assembled into Protofilament Sheets or Bundles To investigate further the oligomeric state of FtsZ under nonassembly conditions, we sedimented protein through glycerol gradients in 0.2 M ammonium bicarbonate. There is no GTPase activity in this buffer, and it does not support assembly with DEAE dextran. Since TmFtsZ shows neither GTPase nor assembly below 50°C, we particularly expected this protein to be monomeric at room temperature. Both Ec- and TmFtsZ sedimented near the position of BSA, consistent with a monomer or dimer. However, rotary shadowed electron micrographs showed a substantial number of short straight protofilaments, mixed with small globular particles that appeared to be 79 monomers (Fig. 5a,b). This association was most striking for TmFtsZ, where the filaments were 5–10 subunits long. This association must have occurred after the centrifugation because these longer filaments would have sedimented well ahead of BSA. Probably the association was inhibited by the hydrostatic pressure in the sedimenting gradient; a similar pressure inhibition was observed for tubulin ring formation [Erickson, 1974]. Remarkably, the assembly of TmFtsZ protofilaments in our ammonium bicarbonate/glycerol buffer was essentially identical regardless of the addition of GTP, GDP, Mg or EDTA. These results show that FtsZ from species as diverse as E. coli and Thermotoga have a tendency to assemble into straight protofilaments, even under conditions that block GTPase and larger assembly. We examined EcFtsZ by negative stain electron microscopy in buffers promoting active GTPase and found long, straight protofilaments in all three buffers (Fig. 5c). Some protofilaments appeared to be single, and others were paired. This assembly required GTP, but not DEAE dextran. Protofilaments were formed at 0°C and persisted during the first few min at 37°C. They disappeared after longer incubation at 37°C, probably when the GTP was largely hydrolyzed. These long protofilaments appear to be the same as those seen by Mukherjee and Lutkenhaus by rotary shadowing , and Wang et al.  by negative stain, and seem to be longer versions of the short filaments in Figure 5a,b. These protofilaments are good candidates for the oligomeric form required for GTPase activity (see under Discussion). Assembly of EcFtsZ protofilaments into larger sheets could be obtained after addition of the polycation DEAE dextran (Fig. 5d). The structure of these sheets and their similarity to the microtubule wall has been described previously [Erickson et al., 1996]. Under some conditions, EcFtsZ plus DEAE dextran produces small tubes, similar to those described by Mukherjee and Lutkenhaus [Mukherjee and Lutkenhaus, 1994]. The structure of these tubes will be described elsewhere. AzFtsZ in MEMK6.5 rapidly formed a white precipitate upon addition of Mg at 0°C. A similar but slower precipitation occurred in PB7.7 after addition of Mg. This precipitation required GTP, and therefore probably represents assembly into meaningful polymers. However, we were not able to visualize these polymers by electron microscopy, perhaps because they are too large. When DEAE dextran was added before the Mg, the visible turbidity was reduced and electron microscopy showed loose protofilament bundles, with a less regular lattice structure than that of EcFtsZ (Fig. 5e). We have yet not found optimal conditions for assembling AzFtsZ. Assembly of TmFtsZ required Mg, GTP, and high temperature, but remarkably did not require DEAE dextran. TmFtsZ in MEMK/Na6.5 plus GTP formed 80 Lu et al. Figure 3. GTPase and Assembly of FtsZ highly ordered protofilament sheets with a very clear resolution of individual protofilaments (Fig. 5f). TmFtsZ sheets were temperature sensitive, and disassembled quickly when the temperature dropped even to 50°C. Fixation by glutaraldehyde was essential to preserve the assembled sheets before preparation of negatively stained specimens. Following assembly and fixation, more than 90% of TmFtsZ could be pelleted by centrifugation at 20,000g, indicating that the assembly was highly efficient. We tested the effect of changing the concentration of DEAE dextran on assembly and GTPase of EcFtsZ. Figure 6a shows that in MEMK6.5 buffer 87% of the FtsZ could be assembled and pelleted by centrifugation with a ratio of 0.6 mg DEAE dextran/FtsZ. The fraction assembled dropped at both lower and higher DEAE dextran, very similar to the assembly of tubulin with DEAE dextran [Erickson and Voter, 1976]. Electron microscopy confirmed that polymers were similar in size at all DEAE dextran concentrations, but much more numerous at 0.6 mg/ml. A reasonable interpretation is that DEAE dextran is stabilizing the polymer by binding electrostatically to and bridging two or more subunits. At lower concentrations of DEAE dextran, assembly is limited by the stoichiometry: 0.06 mg/ml DEAE dextran assembled 0.5 mg FtsZ, compared with 0.42 mg tubulin [Erickson and Voter, 1976]. In excess DEAE dextran, each DEAE dextran molecule may bind a single FtsZ, rather than bridging two or more and stabilizing the polymer. With FtsZ, as with tubulin, the curves were similar at different protein concentrations, i.e., it was the ratio of DEAE dextran to protein that determined the polymerization. Figure 6b shows that GTPase was inhibited by the DEAE dextran-induced polymerization. A complication in this experiment is that the polymer form changed. At 0.06 and 6 mg/ml, the polymers were protofilament sheets, but at the peak assembly, 0.6 mg/ml DEAE dextran, the polymers were tubes. It is likely that the tube polymers are especially inhibited in GTPase, but the protofilament sheets at 0.06 mg/ml are also inhibited Fig. 3. Effect of protein concentration on GTPase activity in TKM7.5 buffer. a: Time course of GTP hydrolysis. Note again that the GTPase activity is indicated by the initial slope of the curves, not by the height of the plateau (which represents the exhaustion of all GTP). The numbers to the right of each curve indicate the protein concentration in µg/ml. The time scale on the x-axis is different for TmFtsZ. b: GTP turnover (moles GTP hydrolyzed per mole of FtsZ per min) in the interval of maximum hydrolysis. At the higher protein concentrations, this was starting from the 0 time point, but at the lower protein concentrations the turnover rate was determined after the lag (points in parentheses). The GTP concentration was 0.15–0.5 mM, and several duplicate experiments indicated no significant difference due to GTP concentration in this range. EcFtsZ and AzFtsZ were assayed at 37°C, TmFtsZ at 70°C. The straight lines fit the points, as well as more complex curves. 81 relative to no DEAE dextran. It is important to remember that FtsZ assembles into single or double protofilaments, which appear to be the oligomer required for the high GTPase, in the absence of DEAE dextran. These individual protofilaments were not pelleted in the centrifugation conditions used. Figure 6b indicates that further assembly of these protofilaments into large sheets or tube polymers substantially inhibits their ability to mediate GTP hydrolysis. DISCUSSION Table I summarizes the GTPase activities of FtsZ determined here and in several independent studies. These hydrolysis rates, 2–10 GTP hydrolyzed per FtsZ per min, are surprisingly high. Most proteins that hydrolyze GTP have a very low intrinsic GTPase, which is accelerated by interaction with other proteins. In a general discussion of the GTPase superfamily, Bourne et al.  comment: ‘‘In many (but not all) GTPases the intrinsic rate constants of GDP release (kdiss.GDP) and GTP hydrolysis (kcat.GTP) are quite low (,0.03 min−1). These are increased by either of two classes of regulatory proteins: the guanine nucleotide release proteins (GNRPs), which catalyze release of bound GDP, promoting its replacement by GTP, and the GTPase activating proteins (GAPs), which speed up GTP hydrolysis.’’ GAPs can accelerate GTP hydrolysis up to 5 orders of magnitude when they bind to their cognate G protein. The mechanism has been revealed most clearly in the recent x-ray structures of Ras●RasGAP [Scheffzek et al., 1997], and Rho●RhoGAP [Rittinger et al., 1997], which show that the GTP molecule is in the interface between the G-protein and the GAP, and makes contact with both proteins. Importantly, most catalytic residues are provided by Ras or Rho, but in both cases one catalytic residue is provided by the GAP. The GAP may also stabilize a conformation of the G-protein favoring the transition state. Tubulin has a very low GTPase activity under most conditions that prevent its assembly into microtubules, but during microtubule assembly each tubulin subunit hydrolyzes one GTP shortly after it enters the microtubule lattice. We have postulated that tubulin acts as its own GAP, stimulating hydrolysis only when two subunits encounter each other for productive assembly [Erickson and O’Brien, 1992]. The GTPase of dynamin has some similarity to that of tubulin. The basal GTP turnover is much higher, about 2 min21 but, like tubulin, it is significantly enhanced to 10 min21 when assembled into polymers [Warnock et al., 1996]. Cooperative assembly onto microtubules or lipid vesicles can increase the activity to 100 min21 [Tuma et al., 1993]. Thus, the 82 Lu et al. Fig. 4. Effect of GTP concentration on GTPase activity of EcFtsZ. The buffer was as in Fig. 3, except for 350 mM KCl (bottom). EcFtsZ was 0.3 mg/ml in all cases. Left, top and bottom, GTP hydrolysis as a function of time at different GTP concentrations, indicated in mM on the right of each curve. Right, maximum GTP turnover, measured after the lag when one exists. GTPase activity of dynamin is expressed primarily when it is assembled into its characteristic polymer, which in this case is a curved protofilament forming rings or stacks of rings [Hinshaw and Schmid, 1995]. An important characteristic of tubulin GTPase is that subunits in the interior of the microtubule cannot exchange nucleotide, so hydrolysis only occurs when new subunits are added to the ends. Since GTP hydrolysis is coupled to assembly, the GTPase is proportional to the concentration of soluble tubulin and the concentration of microtubule ends. GTP hydrolysis can reach 1–3 min–1 per mol of total tubulin during the initial phase of microtubule assembly, but slows to 0.06 min–1 at steady state [O’Brien et al., 1987]. Obviously, steady-state hydrolysis requires a continuous association of subunits onto some microtubules, which must be balanced by an equal dissociation of subunits from other microtubules. This process is known as microtubule dynamic instability [Erickson and O’Brien, 1992]. Continuous assembly and disassembly of dynamin has also been implicated in its GTPase [Warnock et al., 1996], but the mechanism is most clearly established for microtubules. GTPase and Assembly of FtsZ Fig. 5. a,b: Electron microscopy of FtsZ after glycerol gradient sedimentation. Rotary shadowed specimens were prepared at room temperature from gradient fractions in 0.2 M ammonium bicarbonate, 30% glycerol. EcFtsZ (a) shows mostly small globular subunits that we interpret to be FtsZ monomers (arrow), and some dimers and trimers. TmFtsZ (b) shows some monomers (arrow), but mostly short straight protofilaments several subunits in length. c: Single (arrow) and paired (arrowhead) protofilaments imaged by negative stain, from samples of EcFtsZ under conditions of high GTPase. d–f: Protofilament sheets and 83 bundles assembled by EcFtsZ, AzFtsZ, and TmFtsZ. d: EcFtsZ (1 mg/ml) plus 0.6 mg/ml DEAE dextran in MEMK6.5 at 37°C assembles into protofilament sheets. Similar sheets were assembled in TKM7.5 and PB7.7. e: AzFtsZ was assembled with 1 mg/ml protein plus 0.6 mg/ml DEAE dextran at 37°C in PB7.7. It forms bundles of protofilaments in a looser array than EcFtsZ sheets. f: TmFtsZ assembled at 1 mg/ml protein and no DEAE dextran at 70°C in MEMK6.5. This preparation was fixed by adding glutaraldehyde to 1%, and a negatively stained specimen was prepared after 5 min at 70°C. 84 Lu et al. Fig. 6. Assembly and GTPase assay of FtsZ in MEMK6.5 containing 1 mg/ml FtsZ, 1 mM GTP, and DEAE dextran as indicated. a: Assembly of FtsZ at 37°C for 5 min was assayed by centrifugation at 20,000g, 4°C. The pelleted protein was dissolved in 0.1 M NaOH and its concentration determined by the BCA protein assay. The assembly is presented as percentage of FtsZ pelleted. b: GTPase assay of FtsZ under the same conditions. Reactions were assayed every 3 min, and the GTPase activity expressed as turnover of GTP per FtsZ per min. SD were calculated from three (a) or two (b) independent experiments. GTP Hydrolysis Is Coupled to FtsZ Polymerization The sigmoidal dependence of GTPase on FtsZ protein concentration was first observed for FtsZ from B. subtilis [Wang and Lutkenhaus, 1993]. At ,80 µg/ml, the GTPase was essentially zero, but at slightly higher concentrations the specific activity increased very steeply. At 80–150 µg/ml, the GTPase increased from 1.3 to 9.3 min21; then, at 150 µg/ml, the specific activity abruptly plateaued and remained constant at higher concentrations. Wang and Lutkenhaus  concluded that ‘‘the B. subtilis FtsZ must oligomerize in order to express GTPase activity.’’ FtsZ from Haloferax volcanii showed similar behavior at low concentrations: virtually no GTPase at 100 µg/ml, increasing steeply from 0.24 to 1.3 min21 as the concentration increased from 200 to 400 µg/ml [Wang and Lutkenhaus, 1996] (higher concentrations were not explored, so a plateau was not determined). We now report that Ec-, Az-, and TmFtsZ show a concentration dependence of GTPase very similar to BsFtsZ: a low GTPase at ,50 µg/ml protein concentration, increasing sharply from 50 to 300 µg/ml, and a plateau, where the specific GTPase is independent of protein concentration, of .300 µg/ml. The most reasonable explanation is the one suggested previously, that the GTPase activity is expressed only when FtsZ forms an oligomer. The sigmoid shape and steepness of the activation with concentration strongly suggest a nucleated polymerization, in which the oligomer is higher order than a dimer. The abrupt plateau in specific activity at .200–400 µg/ml implies that above this concentration virtually all additional protein associates into oligomers with the same GTPase activity. It is interesting that the average concentration of FtsZ in E. coli was found to be 440 µg/ml, somewhat above this plateau value. The best candidate for the active oligomers are the single and paired protofilaments seen by negative stain electron microscopy (Fig. 5c). These protofilaments were abundant at protein concentrations of .300 µg/ml, and rare at ,50 µg/ml, consistent with the concentration range for formation of the active oligomeric species. The sigmoidal curve implies a multi-subunit nucleated assembly, and this is also suggested by the lag in GTPase seen at low protein concentration, at high GTP, and at 400 mM KCl. However, nucleated assembly would not be expected for single protofilaments. Addition of a subunit to the end of a protofilament should involve the same reaction as formation of a dimer, so assembly of a single protofilament should be an isodesmic reaction with no nucleation phase. It is therefore likely that the active polymers are double protofilaments or larger parallel arrays. Indeed, many double protofilaments are seen in the micrographs, and some images that appear to show single protofilaments may be edge views of double filaments. The electron microscopy so far is suggestive but preliminary, and more will be needed to specify the structure of the active polymers. The Question of Nucleotide Exchange and GTPase Activity of Larger Polymers Continuous GTP hydrolysis requires that GDP be exchanged for GTP following hydrolysis, to initiate a second round. Since hydrolysis appears to be associated with polymers, there are two possible mechanisms for this exchange. One possibility is internal exchange, where the nucleotide would be exchanged from every subunit in the polymer. With microtubules this does not occur; the nucleotide is trapped in the interface between GTPase and Assembly of FtsZ 85 TABLE I. TmFtsZ AzFtsZ EcFtsZ BsFtsZ HvFtsZ Spec. act. Temp. (°C) Protein concn. (mg/ml) KCl (mM) GTP (mM) Lag Buffera Ref.c 4.6 2.2 1.4 0.2 0.05 3.3 2.4 2.3 4.8 7.5 26 0.4–0.7 1.4 1.4 0.6 1.2 2.8 1.3 1.3–9.3 0.2–1.3 90 70 65 50 37 37 37 37 37 37 37 37 37 37 30 37 42 30 37 37 0.8 0.3–0.8 0.8 0.8 0.8 0.3–0.87 0.4–1.5 0.3 0.3 0.3 0.3 0.18–0.27 0.25 1–2 1.0 1.0 1.0 0.4 0.1–0.18 0.2–0.4 50 50 50 50 50 50 50 50 350 50 350 100 200 200 200 200 200 50 100 2000 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 5 5 0.5 5 5 5 5 5 0.5 0.5 0.5 2 2 2 2 2 2 2 2 2 1 1 1 11 1 2 2 2 2 2 2 A A A A A A A A A A A B C C Cb Cb Cb D B E a a a a a a a a a a a b c c d d d e f g aA, 50 mM Tris-HCl, pH 7.5, 50 mM KCl, 5 mM MgAC2 , 0.15–0.5 mM GTP; B, 50 mM Hepes, pH 7.2, 100 mM KCl, 10 mM MgCl2 , 0.5 mM GTP; C, 40 mM Tris-acetate, pH 7.0, 200 mM KAC, 15 mM MgAC2 , 5 mM GTP, 2 mM EDTA, 1 mM DTT, 0.5% Triton X-100; D, 50 mM Tris-HCl, pH 7.5, 50 mM KCl, 2.5 mM MgCl2 , 0.5 mM GTP, 5% glycerol; E, 50 mM Hepes, pH 7.2, 2 M KCl, 10 mM MgCl2 , 0.5 mM GTP. TmFtsZ, Thermotoga maritima FtsZ; EcFtsZ: Escherichia coli FtsZ; AzFtsZ, Azotobacter vinelandii FtsZ; BsFtsZ: Bacillus subtilis FtsZ; HvFtsZ, Haloferax volcanii FtsZ. bFtsZ was preincubated with 5 mM GTP at 37°C for 80 min. ca, this work; b, Fig. 4 of Mukherjee ; c, Fig. 3a of de Boer et al. ; d, Fig. 3b of de Boer et al. ; e, Fig. 2e of Ray Chaudhuri ; f, Fig. 7A of Wang et al. ; g, Fig. 2B of Wang et al. . subunits, so that internal subunits cannot exchange nucleotide. The other possibility is exchange on free subunits accompanied by continuous dissociation and reassociation. With microtubules, nucleotide hydrolysis occurs rapidly after a subunit associates onto the end of a microtubule, and the subunit retains its GDP as long as it remains in the microtubule. During the shortening phase of dynamic instability subunits are released, and once free in solution they can exchange GDP for GTP and be charged for another round of assembly. It is unclear how this scenario might apply to FtsZ, but the similarity of atomic structures [Löwe and Amos, 1998; Nogales et al., 1998] strongly suggests that the nucleotide will be nonexchangeable in the straight protofilament. We observed in Fig. 6 that assembly into larger sheets of protofilaments stabilized by DEAE dextran substantially reduced the GTPase activity. Yu and Margolin  reported recently that 10 mM calcium induced assembly of larger polymers; we have repeated these results (unpublished work) and found that calcium induces assembly of protofilament sheets and tubes that are essentially identical to those induced by DEAE dextran. Yu and Margolin also reported reduction in GTPase at the calcium concentrations giving maximum assembly, similar to our results with DEAE dextran. Apparently the GTP hydrolysis is stimulated primarily by small polymers, the protofilaments shown in Figure 5c, and inhibited when these protofilaments associate into larger polymers. A likely explanation is that dissociation and turnover of subunits is reduced in the larger polymers. A final word of caution is needed concerning the high rate of GTP hydrolysis observed in vitro. Hydrolysis of 2–10 molecules of GTP per min by the 15,000 molecules of FtsZ in a cell would seem a large metabolic burden. Little is known about the dynamics of FtsZ assembly in the cell, so perhaps it is assembled once and maintained in a polymer form that blocks GTP exchange and further hydrolysis. A different enigma is posed by the mutant FtsZ84. This temperature sensitive mutant is defective for cell division at 42°C but can function fully at 30°C. However, its GTPase in vitro is less than one-tenth that of wild-type FtsZ over the entire range of 30–44°C [RayChaudhuri and Park, 1992; de Boer et al., 1992]. This finding suggests that the very high level of GTP hydrolysis observed in vitro may not be biologically important. The lower hydrolysis rates achieved by FtsZ84, or by FtsZ assembled with DEAE dextran may be more relevant to the in vivo function. ACKNOWLEDGMENTS We thank Dr. David Bramhill, Merck, for providing the pET plasmid for expression of EcFtsZ; Dr. Michael Adams, University of Georgia, for Thermotoga DNA; Dr. 86 Lu et al. Paul E. Bishop, North Carolina State University, for Azotobacter DNA; Dr. Limin Zhang, Virginia Polytechnic Institute, for the Azotobacter cosmid library. REFERENCES Addinall, S.G., Bi, E.F., and Lutkenhaus, J. (1996): FtsZ ring formation in fts mutants. J. Bacteriol. 178:3877–3884. Beall, B., Lowe, M., and Lutkenhaus, J. (1988): Cloning and characterization of Bacillus subtilis homologues of Escherichia coli cell division genes ftsA and ftsZ. J. Bacteriol. 170:4855–4864. Bourne, H.R., Sanders, D.A., and McCormick, F. (1991): The GTPase superfamily: Conserved structure and molecular mechanism. Nature 349:117–127. Bramhill, D. and Thompson, C.M. (1994): GTP-dependent polymerization of Escherichia coli FtsZ protein to form tubules. Proc. Natl. Acad. Sci. U.S.A. 91:5813–5817. Brennan, K.J., and Hardeman, E.C. (1993): Quantitative analysis of the human a-skeletal actin gene in transgenic mice. J. Biol. Chem. 268:719–725. Cayley, S., Lewis, B.A., Guttman, H.J., and Record, M.T., Jr. (1991): Characterization of the cytoplasm of Escherichia coli K-12 as a function of external osmolarity: Implications for protein–DNA interactions in vivo. J. Mol. Biol. 222:281–300. Dai, K., and Lutkenhaus, J. (1992): The proper ratio of FtsZ to FtsA is required for cell division to occur in Escherichia coli. J. Bacteriol. 174:6145–6151. de Boer, P., Crossley, R., and Rothfield, L. (1992): The essential bacterial cell division protein FtsZ is a GTPase. Nature 359: 254–256. Donachie, W.D., Begg, K.J., and Vicente, M. (1976): Cell length, cell growth and cell division. Nature 264:328–333. Erickson, H.P. (1974): Assembly of microtubules from preformed, ring-shaped protofilaments and 6-S tubulin. J. Supramol. Struct. 2:393–411. Erickson, H.P. (1997): FtsZ, a tubulin homolog, in prokaryote cell division. Trends Cell Biol. 7:362–367. Erickson, H.P., and O’Brien, E.T. (1992): Microtubule dynamic instability and GTP hydrolysis. Annu. Rev. Biophys. Struct. Biol. 21:145–166. Erickson, H.P., and Stoffler, D. (1996): Protofilaments and rings, two conformations of the tubulin family conserved from bacterial FtsZ to a/b and gamma tubulin. J. Cell Biol. 135:5–8. Erickson, H.P., and Voter, W.A. (1976): Polycation-induced assembly of purified tubulin. Proc. Natl. Acad. Sci. U.S.A. 73:2813–2817. Erickson, H.P., Taylor, D.W., Taylor, K.A., and Bramhill, D. (1996): Bacterial cell division protein FtsZ assembles into protofilament sheets and minirings, structural homologs of tubulin polymers. Proc. Natl. Acad. Sci. U.S.A. 93:519–523. Foster, J.F., and Sterman, M.D. (1956): Conformation change in bovine plasma albumin associated with hydrogen ion and urea binding. II. Hydrogen ion titration curves. Journal of the American Chemical Society 78:3656–3660. Fowler, W.E., and Erickson, H.P. (1979): Trinodular structure of fibrinogen. Confirmation by both shadowing and negative stain electron microscopy. J. Mol. Biol. 134:241–249. Gill, S.C., and Von Hippel, P.H. (1989): Calculation of protein extinction coefficients from amino acid sequence data. Anal. Biochem. 182:319–326. Hinshaw, J.E., and Schmid, S.L. (1995): Dynamin self-assembles into rings suggesting a mechanism for coated vesicle budding [see comments]. Nature 374:190–192. Kubitschek, H.E. (1986): Increase of cell mass during the division cycle of Escherichia coli B/rA. J. Bacteriol. 168:613–618. Löwe, J., and Amos, L.A. (1998): Crystal structure of the bacterial cell-division protein FtsZ. Nature 391:203–206. Ma, X., Ehrhardt, D.W., and Margolin, W. (1996): Colocalization of cell division proteins FtsZ and FtsA to cytoskeletal structures in living Escherichia coli cells by using green fluorescent protein. Proc. Natl. Acad. Sci. U.S.A. 93:12998–13003. Margolin, W., Corbo, J.C., and Long, S.R. (1991): Cloning and characterization of a Rhizobium meliloti homolog of the Escherichia coli cell division gene ftsZ. J. Bacteriol. 173:5822–5830. Margolin, W., and Long, S.R. (1994): Rhizobium meliloti contains a novel 2nd homolog of the cell division gene ftsZ. J. Bacteriol. 176:2033–2043. Mukherjee, A., Dai, K., and Lutkenhaus, J. (1993): Escherichia coli cell division protein FtsZ is a guanine nucleotide binding protein. Proc. Natl. Acad. Sci. U.S.A. 90:1053–1057. Mukherjee, A., and Lutkenhaus, J. (1994): Guanine nucleotidedependent assembly of FtsZ into filaments. J. Bacteriol. 176: 2754–2758. Nogales, E., Wolf, S.G., and Downing, K.H. (1998): Structure of the ab tubulin dimer by electron crystallography. Nature 391: 199–203. O’Brien, E.T., Voter, W.A., and Erickson, H.P. (1987): GTP hydrolysis during microtubule assembly. Biochemistry 26:4148–4156. Ochman, H., Gerber, A.S., and Hartl, D.L. (1988): Genetic applications of an inverse polymerase chain reaction. Genetics 120:621–623. Pla, J., Sanchez, M., Palacios, P., Vicente, M., and Aldea, M. (1991): Preferential cytoplasmic location of FtsZ, a protein essential for Escherichia coli septation. Mol. Microbiol. 5:1681–1686. RayChaudhuri, D., and Park, J.T. (1992): Escherichia coli cell-division gene ftsZ encodes a novel GTP-binding protein. Nature 359: 251–254. Rittinger, K., Walker, P.A., Eccleston, J.F., Smerdon, S.J., and Gamblin, S.J. (1997): Structure at 1.65 Å of RhoA and its GTPaseactivating protein in complex with a transition-state analogue. Nature 389:758–761. Scheffzek, K., Ahmadian, M.R., Kabsch, W., Wiesmüller, L., Lautwein, A., Schmitz, F., and Wittinghofer, A. (1997): The Ras–RasGAP complex: structural basis for GTPase activation and its loss in oncogenic ras mutants. Science 277:333–345. Smith, P.K., Krohn, R.I., Hermanson, G.T., Mallia, A.K., Gartner, F.H., Provenzano, M.D., Fujimoto, E.K., Goeke, N.M., Olson, B.J., and Klink, D.C. (1986): Measurement of protein using bicinchonic acid. Anal. Biochem. 150:76–85. Trueba, F.J. and Woldringh, C.L. (1980): Changes in cell diameter during the division cycle of Escherichia coli. J. Bacteriol. 142:869–878. Tuma, P.L., Stachniak, M.C., and Collins, C.A. (1993): Activation of dynamin GTPase by acidic phospholipids and endogenous rat brain vesicles. J. Biol. Chem. 268:17240–17246. Wang, X., and Lutkenhaus, J. (1993): The FtsZ protein of Bacillus subtilis is localized at the division site and has GTPase activity that is dependent upon FtsZ concentration. Mol. Microbiol. 9:435–442. Wang, X., and Lutkenhaus, J. (1996): FtsZ ring: The eubacterial division apparatus conserved in archaebacteria. Mol. Microbiol. 21:313–319. Wang, X.D., Huang, J.A., Mukherjee, A., Cao, C., and Lutkenhaus, J. (1997): Analysis of the interaction of FtsZ with itself, GTP, and FtsA. J. Bacteriol. 179:5551–5559. Warnock, D.E., Hinshaw, J.E., and Schmid, S.L. (1996): Dynamin self-assembly stimulates its GTPase activity. J. Biol. Chem. 271:22310–22314. Yu, X.C., and Margolin, W. (1997): Ca21-mediated GTP-dependent dynamic assembly of bacterial cell division protein FtsZ into asters and polymer networks in vitro. EMBO J. 16:5455–5463.