A set of ftsZ mutants blocked at different stages of cell division in Caulobacter



FtsZ is required throughout the cell division process in eubacteria and in archaea. We report the isolation of novel mutants of the FtsZ gene in Caulobacter crescentus. Clusters of charged amino acids were changed to alanine to minimize mutations that affect protein folding. Molecular modelling indicated that all the clustered-charged-to-alanine mutations had altered amino acids at the surface of the protein. Of 13 such mutants, four were recessive-lethal, three were dominant-lethal, and six had no discernible phenotype. An FtsZ depletion strain of Caulobacter was constructed to analyse the phenotype of the recessive-lethal mutations and used to show that they blocked cell division at distinct stages. One mutation blocked the initiation of cell division, two mutations blocked cell division randomly, and one mutation blocked both early and late stages of cell division. The effect of the recessive mutations on the subcellular localization of FtsZ was determined. Models to explain the various mutant phenotypes are discussed. This is the first set of recessive alleles of ftsZ blocked at different stages of cell division.


Bacterial cell division is a complex process that requires at least nine genes in Escherichia coli, ftsZ, A, I, K, L, N, Q, W and ZipA (Rothfield et al., 1999). Of these genes, ftsZ acts at the earliest known stage of cell division. FtsZ null mutants are unable to initiate cell division and form long smooth filaments. FtsZ localizes to the site of cell division before any visible cellular constriction and forms a ring structure called the Z ring (Bi and Lutkenhaus, 1991). The Z ring stays at the leading edge of the septum during invagination and disassembles once septation is completed (Bi and Lutkenhaus, 1991; Addinall et al., 1996). FtsZ binds and hydrolyses GTP in vitro (de Boer et al., 1992; RayChaudhuri and Park, 1992; Mukherjee et al., 1993) and can polymerize in vitro into protofilaments and sheets in a GTP-dependent manner (Bramhill and Thompson, 1994; Erickson et al., 1996; Yu and Margolin, 1997; Lu et al., 1998; Mukherjee and Lutkenhaus, 1998). It has been proposed that GTP-bound FtsZ polymerizes into a ring structure during cell division, and that GTP hydrolysis leads to the instability and depolymerization of the Z ring (Lutkenhaus and Addinall, 1997). FtsZ is required for the recruitment of other cell division proteins to the division site. FtsA, I, K, L, Q, N, W and ZipA have all been shown to localize to the septum in an FtsZ-dependent manner (Ma et al., 1996; Addinall et al., 1997a; Buddelmeijer et al., 1998; Wang et al., 1998; Yu et al., 1998; Chen et al., 1999; Ghigo et al., 1999; Hale and De Boer, 1999; Liu et al., 1999; Weiss et al., 1999).

FtsZ can be divided into three regions based on sequence similarity between FtsZ from different bacteria: a long N-terminal conserved region (NTR) of approximately 300 amino acids that is absolutely conserved; a spacer region (spacer) that varies in both length and sequence; and a short C-terminal conserved region (CTR) that is present in many but not all FtsZ proteins (Fig. 1). The NTR region is required for FtsZ–FtsZ interaction and polymerization, whereas the CTR region is required for FtsZ–FtsA interaction (Ma et al., 1996; 1997; Wang et al., 1997; Din et al., 1998). The NTR region can be subdivided into two domains based on the crystal structure of Methanococcus jannaschii FtsZ: an N-terminal GTPase domain and a C-terminal domain of unknown function (referred to as the NC domain), which are bridged by a central helix, H7 (Löwe and Amos, 1998). The GTP-binding domain is similar to that of α/β tubulin (Löwe and Amos, 1998; Nogales et al., 1998a).

Figure 1.

Alignment of FtsZ proteins and sites of mutations. Primary sequence of FtsZ from four different bacteria (in descending order: C. crescentus, M. jannaschii, Bacillus subtilis and E. coli) are aligned. Mutation sites are underlined, and bold residues were changed to alanines except for mutant 109, which has a G to S change (**). Structural elements are indicated and numbered sequentially as H0–H10 and S1–S10 for helices and strands, respectively, as described previously (Nogales et al., 1998b). Loops with residues involved in direct contact with the nucleotide are marked T1–T7 (Nogales et al., 1998b).

FtsZ is required throughout cell division (Addinall et al., 1997b). To study the function of FtsZ at different stages of the cell division process, we wanted to identify FtsZ mutants that are blocked at various stages. In this paper, we describe new mutants of FtsZ obtained by clustered-charged-to-alanine scanning mutagenesis of Caulobacter FtsZ (Wertman et al., 1992; Reijo et al., 1994). Clustered-charged-to-alanine mutagenesis has been used successfully to identify important surface residues in yeast tubulin and actin (Wertman et al., 1992; Reijo et al., 1994). This method is based on the observation that charged residues tend to be at the surface of a protein, and that changes in surface residues are not usually disruptive to protein structure. Because residues involved in protein–protein interaction and enzymatic activity are located at the surface of proteins, this method enriches for mutations that disrupt protein activity but not structure. Caulobacter crescentus is a powerful experimental system for studying cell division because of the ease with which synchronized populations can be obtained (Brun et al., 1994; Brun and Janakiraman, 2000). Each cell division gives rise to two morphologically different progeny cells: a motile swarmer cell and a sessile stalked cell. FtsZ, which is absent from swarmer cells at the beginning of the cell cycle (Quardokus et al., 1996), begins to accumulate during swarmer-to-stalked cell differentiation and forms a Z ring at the future site of cell division (Kelly et al., 1998). At the end of the cell cycle, FtsZ becomes unstable and is rapidly degraded (Kelly et al., 1998). The inherent instability of Caulobacter FtsZ provides an important tool for replacing wild-type FtsZ with mutant forms of the protein.

Here, we describe the construction and properties of a FtsZ depletion strain of Caulobacter used to facilitate the analysis of ftsZ mutants. Out of 14 mutants studied, four mutants were dominant-lethal and four mutants were recessive to wild-type FtsZ. These mutants were blocked at different stages of cell division. The remaining six mutants had a wild-type phenotype. A molecular model of Caulobacter FtsZ was used in conjunction with the mutant phenotypes and immunolocalization data to propose a basis for the defects in the various mutants.


Construction and characterization of an FtsZ depletion strain

To facilitate the study of ftsZ in Caulobacter, we constructed strain YB1585 (Fig. 2A), in which the chromosomal wild-type ftsZ is controlled by the xylose-inducible xylX promoter (see Experimental procedures). The YB1585 strain also has a tandem duplication of the first 163 codons of ftsZ transcribed by the native ftsZ promoter (Fig. 2A). A similar construct of FtsZ gave no phenotype when expressed from a high-copy plasmid, demonstrating that this truncated FtsZ does not interfere with cell division (Din et al., 1998). Furthermore, the resulting truncated FtsZ fragment cannot be detected on a Western blot (not shown), indicating that the truncated gene is probably a null allele. YB1585 required xylose for growth on plates (Fig. 2B) and for cell division in PYE liquid medium (Fig. 3A and B). The requirement for xylose could be complemented by the low-copy plasmid plac290ftsZ, which contains the ftsZ gene, but not by the vector pRKlac290 alone (Figs 2B and 3C). Microscopic examination of a culture at different times after transfer of exponentially growing YB1585 cells from PYE-xylose medium to PYE-glucose medium indicated that cells began to filament 1 h after the transfer (Fig. 3D) and continued to elongate without initiating cell division (Fig. 3E and F). The cell division defect of YB1585 in PYE-glucose was fully complemented by a copy of ftsZ on plasmid plac290ftsZ (Fig. 3C). Immunoblotting of cell extracts from PYE-xylose cultures showed that the FtsZ concentration of exponentially growing YB1585 was similar to that of a wild-type strain (Fig. 4A). The concentration of FtsZ decreased rapidly after the removal of xylose. After 30 min, the FtsZ concentration had decreased to ≈ 12% of the original concentration, as determined by densitometry, and only 6% remained after 2 h (Fig. 4A). These results indicate that the xylose-dependent cell growth and cell division of YB1585 results from the control of ftsZ expression and that FtsZ is required for the initiation of cell division in Caulobacter. Furthermore, wild-type FtsZ can be rapidly depleted from YB1585 by the removal of xylose. The rapid disappearance of FtsZ upon shift of YB1585 to PYE-glucose medium was expected, as most of the FtsZ in wild-type cells is rapidly degraded at the end of each cell cycle (Quardokus et al., 1996; 2001; Kelly et al., 1998).

Figure 2.

The FtsZ depletion strain YB1585.

A. YB1585 was created by integrating plasmid pBJM at the ftsZ locus by homologous recombination. The C-terminal truncated ftsZ (ftsZ163ΔC) is a null allele, and the wild-type ftsZ allele is under the control of the xylX promoter. In the absence of xylose, wild-type ftsZ is not expressed.

B. NA1000, YB1585, YB1585/pRKLac290 (control plasmid) and YB1585/plac290ftsZ were streaked onto PYE plates with or without xylose as indicated. The NA1000 sectors were from different plates processed in the same way.

Figure 3.

Phenotype of YB1585 grown in PYE-xylose or PYE-glucose medium.

A. YB1585 strain grown overnight in PYE-xylose medium.

B. YB1585 strain grown overnight in PYE-glucose medium.

C. YB1585/plac290ftsZ grown overnight in PYE-glucose medium.

D–F. Exponentially growing YB1585 cells were taken from PYE-xylose medium, washed with PYE-glucose medium, then inoculated into PYE-glucose medium at 1:200 dilution. Pictures were taken at 1 h (D), 2 h (E) and 4 h (F) after the transfer.

Figure 4.

Immunoblot of YB1585 and YB1585 containing various mutant FtsZ using an anti-FtsZ antibody.

A. YB1585 cells from exponential phase were transferred from PYE-xylose to PYE-glucose medium. Samples were taken at the times indicated after the transfer and processed for immunoblotting.

B. Swarmer cells and stalked cells were taken from exponential cultures of wild-type strain NA1000 or YB1585 and processed for immunoblotting.

C. Immunoblot of YB1585 strain containing recessive mutants. Equal amounts of cells were loaded into each lane. Wild-type NA1000 grown in PYE medium (lane 1), YB1585 (lanes 2 and 3), YB1585/plac290ftsZ (lanes 4 and 5), YB1585/plac290-86 (lanes 6 and 7), YB1585/plac290-142 (lanes 8 and 9), YB1585/plac290-198 (lanes 10 and11) and YB1585/plac290-254 (lanes 12 and 13) were grown in PYE-xylose medium and transferred to either PYE-xylose medium (+) or PYE-glucose medium (–) as indicated above the lanes. Cell samples were taken at 7 h after the transfer.

We examined the cell cycle progression of YB1585 by isolating swarmer cells from a PYE-xylose culture and resuspending them into PYE-xylose and PYE-glucose medium. Cells in the PYE-xylose medium differentiated and progressed at a wild-type pace through the cell division cycle. By 90 min into the cell cycle, xylose-grown YB1585 cells had initiated cell division (data not shown), and they had deep constrictions by 105 min (data not shown). Synchronized YB1585 cells grown in PYE-glucose never initiated cell division and grew as long filaments (results not shown). We compared the concentration of FtsZ in swarmer and stalked cells of YB1585 and wild-type strain NA1000. In NA1000, only stalked cells contained FtsZ (Fig. 4B), as reported previously (Quardokus et al., 1996). Swarmer cells of YB1585 contained a low concentration of FtsZ compared with stalked cells (Fig. 4B), indicating that FtsZ is mostly depleted from swarmer cells at every cell cycle. This is consistent with the fact that proteolysis is the major determinant of the cell cycle variation in FtsZ concentration (Kelly et al., 1998). These results indicate that YB1585 behaves essentially as wild type with respect to cell division and FtsZ regulation when grown in xylose medium. The rapid degradation of FtsZ in swarmer cells indicates that wild-type FtsZ can be replaced by a mutant form of FtsZ within one cell cycle if YB1585 is grown in glucose in the presence of an expressed mutant allele.

Design and analysis of the mutations

In an effort to obtain FtsZ mutants that are blocked at different stages of cell division, we used clustered-charged-to-alanine site-directed mutagenesis to mutagenize surface residues. We targeted clusters of five adjacent residues containing two or more charged residues and changed them to alanines to ensure minimal disruption of protein structure. There are 53 such clusters in FtsZ; 31 are in the NTR region of the protein and are more likely to be involved in conserved functions of FtsZ. We mutated 12 clusters in the NTR and one cluster in the spacer (Fig. 1). In addition, we mutated glycine 109 to serine because the same amino acid substitution gave rise to a temperature-sensitive allele of E. coli ftsZ, ftsZ84, with altered GTP binding and hydrolysis. After the crystal structure of M. jannaschii was published (Löwe and Amos, 1998), we constructed a three-dimensional model of the Caulobacter FtsZ NTR region based on its homology to M. jannaschii FtsZ using molecular modelling. The location of the mutations is shown on the model of Caulobacter FtsZ (Fig. 5). All 12 mutated clusters in the NTR region are fully or partially exposed at the protein surface, except G109S, which is predicted to lie in the GDP-binding pocket.

Figure 5.

Model of the Caulobacter FtsZ three-dimensional structure. The N-terminal 330 amino acids were modelled based on the crystal structure of M. jannaschii FtsZ. All the secondary structure backbones (helix and sheets) were the same as M. jannashcii FtsZ except that there is no H0 in Caulobacter FtsZ.

A. Ribbon diagram of the Caulobacter FtsZ modelled structure showing two views of the structure rotated 180° vertically. The mutated residues are shown in ball and stick form, and GDP is shown in space-filled form. Helix H7 and loop T7 are indicated.

B. Surface diagram of Caulobacter FtsZ modelled structure. The mutated residues are shown as dark.

To study the effect of the mutations, individual ftsZ mutants were cloned into the low-copy-number plasmid pRKlac290 downstream from the native ftsZ promoter, and the resulting plasmids were conjugated into YB1585. To determine the extent to which the mutant FtsZ can support cell division, transconjugants containing the individual mutants were grown in PYE-xylose medium and then transferred to PYE-glucose medium to shut down the expression of wild-type FtsZ from the chromosome. The cells were examined after 7 h of growth in PYE-glucose medium. In the presence of xylose, both the chromosomal wild-type ftsZ and the plasmid-borne ftsZ are expressed (Fig. 4C). This allows growth of the strains if they contain a recessive-lethal mutation in the plasmid-borne copy of ftsZ, but not if the mutants are dominant. In the absence of xylose, the chromosomal Pxyl–ftsZ allele was not expressed (Fig. 4C, lane 3), and only the plasmid-borne ftsZ, controlled by the native ftsZ promoter, was produced (Fig. 4C, lane 5). The concentration of FtsZ in YB1585/plac290ftsZ grown in PYE-glucose or in PYE-xylose was similar to that in wild-type strain NA1000 (Fig. 4C, lanes 1, 4 and 5). All the mutants were tested for possible temperature-sensitive phenotypes by growing them at different temperatures. Except for mutant 142 (see below), all the mutants had similar phenotypes at 22°C, 30°C and 37°C.

The mutants were grouped into three classes based on their phenotypes in YB1585 (Table 1). One class of mutants, mutants 59, 125, 178, 291, 303 and 441, had a phenotype indistinguishable from YB1585/plac290ftsZ in PYE-glucose medium, indicating that the mutated residues are not critical for cell division (data not shown). Another class of mutants comprised mutants that were unable to support cell division in PYE-glucose medium, mutants 86, 142, 198 and 254 (Fig. 6D, F, H and J). Because cells had some problems dividing properly when these mutants were expressed together with wild-type FtsZ in PYE-xylose (Fig. 6C, E, G and I), these mutations have a slight dominant effect and are thus not truly recessive. Nonetheless, we will refer to them as recessive mutants in the text to distinguish them from the dominant-lethal mutants. Immunoblotting confirmed that the level of FtsZ produced by the recessive mutants was similar to that of wild-type strain NA1000, suggesting that the stability of FtsZ was not changed by the mutations (Fig. 4C, lanes 7, 9, 11 and 13). Mutant 86 migrated more rapidly than wild-type FtsZ and other mutant FtsZ on an SDS–PAGE gel (Fig. 4C, lanes 6 and 7). Sequencing of the mutant gene confirmed that there were no other sequence changes besides the site-directed mutagenized nucleotides (not shown). Conveniently, the immunoblot of YB1585/plac290-86 clearly establishes that the mutant FtsZ is the only remaining form of FtsZ after growth in PYE-glucose for 7 h (Fig. 4C, lane 7). YB1585/plac290-86 grown in xylose contained two- to threefold more mutant FtsZ than wild-type FtsZ and contained much less wild-type FtsZ than YB1585 by itself. As the concentration of FtsZ is mostly controlled by proteolysis, the relative amounts of wild-type and mutant FtsZ in YB1585/plac290-86 probably reflect the relative copy number of the two alleles. We were unable to obtain transconjugants in YB1585 for the last class of mutants (mutants 109, 216, 237 and 278), suggesting that these mutants are dominant-lethal.

Table 1. Phenotype of ftsZ mutants.
MutantLocationPhenotype in YB1585Overproduction phenotype
Wild-type alleles
 WT FtsZ Wild typeMultiple mid-cell constrictions
 DRR59AAAS3Wild typeMultiple mid-cell constrictions
 RER125AAAH4Wild typeMultiple mid-cell constrictions
 ER178AAH6-H7Wild-typeMultiple mid-cell constrictions
 DPE291APAH10-S9Wild typeFilamentous
 DPSLE303APSLAS9-S10Wild typeMultiple mid-cell constrictions
 EERR441AAAASpacerWild typeMultiple mid-cell constrictions
Recessive-lethal alleles
 EE86AAH3Filamentous with constrictionsND
 EGR142AGAH5Deep and extended constriction with
 extended filamentation
Deep and extended constriction
 RSITD198ASITAH7Filamentous with side knobsFilamentous and short areas of deep constriction
 DE254AAH9-S8Filamentous with constrictionsND
Dominant-lethal alleles
 G109SS4-H4Dominant-lethalDeep and extended constriction
 DVR216AVAH8Dominant-lethalCoiled and filamentous. Slightly thinner in the middle
 EDR237AAAS7-H9Dominant-lethalFilamentous with tight twists
 EVDE278AVAAH10Dominant-lethalFilamentous and short areas of deep constriction
Figure 6.

Phenotype of the recessive mutants in YB1585. YB1585/plac290ftsZ (A and B), YB1585/plac290-198 (C and D), YB1585/plac290-86 (E and F), YB1585/plac290-254 (G and H), YB1585/plac290-142 (I and J) were grown in PYE-xylose and transferred to either PYE-xylose (A, C, E, G and I) or PYE-glucose (B, D, F, H and J) at 30°C. Pictures were taken 8.5 h after the transfer.

K and L. YB1585/plac290-142 was transferred from PYE-xylose to PYE-glucose and grown at 22°C for 8.5 and 18 h respectively.

Characterization of the recessive mutants

The four recessive mutants exhibited distinct cell division phenotypes when grown in PYE-glucose (Fig. 6). PYE-glucose cultures of mutant 198 consisted of two kinds of cells. Approximately 45% of the cells were filamentous, whereas the rest of the population was composed of shorter cells, some of which were similar in morphology to wild-type cells. The ability of some cells to divide may result from a small amount of wild-type FtsZ produced by the low-level transcription of the Pxyl promoter that occurs even in the absence of xylose (Meisenzahl et al., 1997). This small amount of FtsZ may be able to provide the nucleation required to initiate polymerization in some cells. Filamentous cells of mutant 198 were blocked at the earliest stage of the four recessive mutants and had no signs of constriction (Fig. 6D). The filaments contained small areas of outgrowth on their sides (Fig. 6D). Occasionally, a stalk could be seen at the end of the outgrowth area (not shown). We used an affinity-purified anti-FtsZ antibody to immunolocalize FtsZ in mutant 198 grown in PYE-glucose. Shorter constricting cells had clear Z rings at the constriction site (Fig. 7A). However, no Z rings were seen in the filamentous cells; only a diffuse FtsZ staining was seen throughout the filaments (Fig. 7A). This indicates that the early cell division block of mutant 198 is likely to be caused by the inability of the mutant FtsZ to localize to the site of cell division or by an inability to polymerize to form a Z ring.

Figure 7.

Immunolocalization of FtsZ in YB1585 containing mutants 198, 86, 254 and 142, respectively, in plac290 (A–D) or in NA1000 expressing mutant 109 from plasmid pUJ142 (E).

A–C. Cells were grown in PYE-xylose and transferred to PYE-glucose medium to grow for 7 h (A and C) or overnight (B).

D. Cells were grown in PYE-xylose at 30°C and transferred to PYE-glucose medium to grow overnight at room temperature.

E. NA1000/pUJ142-109 was grown in PYE-glucose and then induced to express mutant 109 by adding xylose for 5 h. FtsZ was visualized by immunofluorescence microscopy using an affinity-purified antibody against FtsZ and FITC-conjugated secondary antibody. For each blot, from left to right are phase-contrast images, immunofluorescence images and an overlay of phase-contrast/immunofluorescence images of the same field. All images are at the same magnification.

In mutant 86, over 90% of the cells exhibited cell division defects and were filamentous (Fig. 6F). Constrictions could be seen in ≈ 70% of the filaments. The depth of constrictions varied, and there were many sites of deep constriction. Immunolocalization of FtsZ in mutant 86 showed that Z rings were present at ≈ 60% of the constrictions of filamentous cells. Many of the constriction sites in mutant 86 had no FtsZ staining, suggesting that the Z ring may have disassembled prematurely. Z rings were also present at many sites where no constrictions were visible (Fig. 7B). The Z rings seen at sites without constriction could be functional rings that have not yet begun to constrict. The abundance of these unconstricted rings suggests that mutant 86 forms Z rings that constrict slowly or often forms non-functional Z rings. In addition, filaments of mutant 86 had extended regions without Z rings, suggesting that it may also have problems with Z ring assembly.

Mutant 254 had many of the characteristics of mutant 86; over 90% of the cells exhibited cell division defects, and more than 90% of the filamentous cells had constrictions (Fig. 6H). When grown in PYE-glucose, mutant 254 also had many cells with a minicell constriction near the cell end (data not shown). Z rings were often localized to the sites of cell constriction (Fig. 7C). FtsZ staining was often seen as an extended band, and these bands were not always associated with constrictions; it is not known whether these staining patterns are indicative of adjacent Z rings.

Mutant 142 was the only mutant that exhibited a temperature-sensitive phenotype. At 30°C, most of the cells were filamentous with no constrictions (Fig. 6J), and 10% of the cells had extended areas of pronounced constriction (data not shown). At 22°C, most of the cells had extended areas of pronounced constriction (Fig. 6K). Both at 22°C and 30°C, the constrictions of mutant 142 were deeper than most of the constrictions of mutants 86 and 254. One striking feature of mutant 142 at 22°C was the simultaneous presence of extended areas of constriction and of extended areas with no constriction. Surprisingly, FtsZ staining could barely be detected in the deeply constricted areas of mutant 142, and relatively strong FtsZ staining without rings or bands could be detected in the unconstricted areas (Fig. 7D). To ensure that our inability to detect FtsZ in the deeply constricted regions of mutant 142 was not caused by the depth of the constrictions, we also examined a wild-type strain expressing wild-type FtsZ, which caused deep constrictions in extended areas as described previously (Din et al., 1998). FtsZ staining was clearly visible in the deeply constricted areas (data not shown). These results suggest that mutant 142 was able to support cell division until a very late stage and that the Z ring formed by mutant 142 disassembles before the completion of cell division.

Characterizartion of the dominant-lethal mutants

To study the phenotype of the dominant-lethal mutants, we cloned them in vector pUJ142, replacing the native ftsZ promoter with the xylX promoter. The resulting plasmids were conjugated into wild-type strain NA1000 in the absence of xylose to prevent the expression of the mutant proteins. To analyse the defects of the dominant-lethal mutants, we compared the effect of their expression from plasmid pUJ142 with that of wild-type FtsZ from the same plasmid. Xylose induction of wild-type FtsZ expression delayed cell separation, and multiple constrictions appeared at mid-cell within 2 h as described previously (Din et al., 1998). Cell separation eventually occurred and was very apparent after 5 h of induction (Fig. 8B). Expression of the four dominant-lethal mutants inhibited cell division at different stages (Fig. 8C–F). Western blot analysis showed that the expression level of the dominant-lethal mutants was comparable with that of wild-type FtsZ from the same plasmid (data not shown). This suggests that the dominant phenotype of these mutants was not the result of altered protein stability.

Figure 8.

Phenotypes of wild-type NA1000 Caulobacter overexpressing dominant-lethal FtsZ mutants.

A. NA1000/pUJ142ftsZ grown in PYE-glucose medium.

B. NA1000/pUJ142ftsZ grown in PYE-glucose medium and transferred to PYE-xylose medium for 5 h.

C. NA1000/pUJ142-109 grown in PYE-glucose medium and transferred to PYE-xylose medium for 5 h.

D. NA1000/pUJ142-216.

E. NA1000/pUJ142-237 grown in PYE-glucose medium and transferred to PYE-xylose medium for 5 h.

F. NA1000/pUJ142-278 were grown in PYE-glucose medium and transferred to PYE-xylose medium for 5 h.

All pictures are at the same magnification.

G–J. TEM pictures of wild-type NA1000 (G), NA1000/pUJ142-109 (H), NA1000/pUJ142-142 (I) and NA1000/pUJ142-278 (J) grown in PYE-xylose medium for 5 h. The bar represents 1 µm in (G) and (I) and 500 µm in (J) and (H).

Mutant 237 produced filamentous cells without any constrictions (Fig. 8E). However, the filaments produced by mutant 237 were not as smooth as those produced by the FtsZ depletion strain YB1585; they had many kinks over their surface. Mutant 216 produced highly coiled smooth filaments (Fig. 8D). These results indicate that mutants 216 and 237 inhibit the initiation of cell division when they are overexpressed. Immunofluorescence microscopy revealed strong foci of FtsZ staining throughout the cells (not shown).

The expression of mutants 278 and 109 produced extended areas of deep constriction at the mid-cell, ≈ 100 nm thick compared with a normal cell width of 600–700 nm (Fig. 8C, F, H and J). Examination of cells expressing mutant 278 early after induction revealed that all the cells could begin constriction and were blocked at a very late stage of constriction (not shown). For mutant 109, > 95% of the cells had extended areas of deep constriction, and the length of the constricted areas was substantially longer than for mutant 278 (compare Fig. 8C and F). Immunofluorescence of cells overexpressing mutant 109 revealed that these cells had almost all their FtsZ staining confined to the constricted area, with very little FtsZ at the unconstricted cell ends (Fig. 7E). We chose a rare field in which a seemingly wild-type cell containing a Z ring was present for comparison (Fig. 7E).

Mutant 109 has the same mutation as the temperature-sensitive ftsZ84 allele of E. coli. This mutation greatly reduces the GTP binding and GTPase activity of E. coli FtsZ at all temperatures. Mutants 109 and 142 both have mutations in residues that are predicted by the FtsZ crystal structure to be in direct contact with GDP (Fig. 5) (Löwe and Amos, 1998). Therefore, we tested whether the overexpression of mutant 142 conferred a similar phenotype to that of mutant 109 and found that it did; it produced cells with extended areas of deep constrictions (Fig. 8I).


Out of 14 mutants constructed in this study, four were recessive-lethal, four were dominant-lethal, and the remaining six had a wild-type phenotype. Interestingly, the non-wild-type mutants were blocked at different stages of cell division. After construction of our mutants, the three-dimensional structure of M. jannaschii FtsZ was published, allowing us to determine the position of our mutations on the FtsZ structure. Molecular modelling of Caulobacter FtsZ indicated that all our clustered-charged-to-alanine mutants had mutagenized residues exposed at the surface of the protein (Fig. 6). This high yield of surface mutations using clustered-charged-to-alanine mutagenesis is similar to that obtained for yeast actin (Wertman et al., 1992). This study also reports the construction and characterization of a depletion strain of FtsZ in which ftsZ expression is controlled by a xylose-inducible promoter. We have shown previously that expression of some dominant-lethal alleles of FtsZ prevented the initiation of cell division, suggesting that Caulobacter FtsZ is required for the initiation of cell division as shown in other bacteria (Din et al., 1998). The phenotype of FtsZ depletion presented here confirms this hypothesis. In the presence of xylose, this strain expresses FtsZ constitutively at a concentration similar to that of a wild-type strain. Because the control of FtsZ level is mainly exerted by FtsZ degradation (Kelly et al., 1998), swarmer cells were mostly depleted of FtsZ. This allows the rapid replacement of wild-type FtsZ with mutant forms.

Recessive alleles

The four recessive alleles constructed in this study had three distinct phenotypes. Mutant 198 was unable to initiate cell division and had small surface protrusions. Mutants 86 and 254 could initiate cell division, but not complete it, and had constrictions of varying depth. Mutant 142 had extended regions of both deep constriction and regions without constriction. None of the recessive mutations had a phenotype identical to the depletion of FtsZ, indicating that they are not null alleles. Furthermore, all the proteins were expressed at the same level as wild-type FtsZ, indicating that they were stable and likely to be correctly folded.

The early block of mutant 198 can be explained by its inability to form Z rings, which could be caused by a total lack of function, by an inability to localize or by an inability to polymerize. We think that mutant 198 is still partly functional based on several lines of evidence. Cells expressing only mutant 198 had many small surface protrusions along the filaments, suggesting that its expression had effects on the structure of the cell wall. Cells expressing both mutant 198 and wild-type FtsZ were slightly filamentous, indicating that mutant 198 had a slight dominant effect. Finally, when mutant 198 was expressed at a higher level with wild-type FtsZ, cells were mostly filamentous with short areas of deep constriction (Table 1). Mutation 198 is located near the end of the central helix (H7) linking the GTPase domain and the NC domain of the NTR region. This position is far from the GTP-binding pocket, indicating that the mutated residues are not directly involved in GTP binding, which is required for polymerization.

Mutants 86 and 254 were blocked after the initiation of cell division. The fact that the constrictions in these mutants were of different depth indicates that cell division could arrest at any stage. In wild-type cells, Z rings can be detected until the very last stages of cell division (Quardokus et al., 2001). As FtsZ was not localized at many of the constriction sites in mutants 86 and 254, it is likely that the Z rings produced by these mutants are unstable. This is consistent with the random division arrest. The instability of the Z ring could result from a deficiency in interaction with proteins required for stabilizing the Z ring, or FtsZ–FtsZ interaction could be impaired. Loss-of-function mutants of several cell division proteins, such as FtsA, I, K and Q in E. coli, also produce partially constricted filamentous cells (Addinall et al., 1996; Yu et al., 1998). These proteins may be required for the progression of ring constriction or stabilization of the Z ring. Mutants 86 and 254 may be deficient in interacting with some of these proteins. In addition, these mutants, especially mutant 86, may form non-functional Z rings, as many rings were present at non-constricting sites.

Mutant 142 was the only one of our mutants that exhibited a temperature-sensitive phenotype in the range tested (22–37°C). At 30°C, most of the cells were smooth and filamentous, indicating that mutant 142 is unable to initiate cell division at 30°C, but a small proportion of cells had deep constrictions. At 22°C, most of the cells had deep extended constrictions and extended regions with no constriction, whereas a small percentage of cells were filamentous without any constriction. The deep constrictions in mutant 142 at 22°C indicate that mutant 142 forms Z rings that can support cell division through most of the process but cannot support the final stages of constriction. FtsZ was essentially absent from the constricted areas, indicating that the Z rings formed by mutant 142 are destabilized at the end of cell division. The fact that FtsZ was evenly distributed in the long unconstricted areas of the cells suggests that, by itself, mutant 142 is not able to initiate cell division at 22°C. We hypothesize that cell division was initiated early in the depletion process by wild-type FtsZ. If this is the case, it implies that, once localization and polymerization begin as a result of the presence of wild-type FtsZ, mutant 142 can continue to polymerize on that template to form extended areas of constriction and is completely functional until the late stages of cell division. This is supported by the fact that, when mutant 142 was overexpressed in a wild-type strain, all the cells had extended areas of deep constriction and no extended regions without constrictions. Therefore, it appears that mutation 142 inhibits the ability of FtsZ to both initiate and complete division.

One of the residues mutated in mutant 142 is a highly conserved glutamate at position 142, which is identical in all the 51 ftsZ sequences we checked. The crystal structure of FtsZ indicates that residue E142 is in direct contact with the nucleotide ribose (Löwe and Amos, 1998). Therefore, it is likely that mutant 142 has altered GTP-binding and/or hydrolysis activities. The fact that cells from mutant 142 were able to form deep constrictions suggests that mutant 142 is able to polymerize. Amino acid R144 was also mutated to alanine in mutant 142. R144 has not been implicated in GTP binding or hydrolysis, and our molecular model indicates that it protrudes out of the surface of the protein at the edge of the GTP binding site. Construction of each single mutant will be required to distinguish between the effect of the two individual mutations.

Dominant alleles

One important consideration in interpreting the dominant phenotypes is that the wild-type allele of FtsZ was expressed from the chromosome at the same time as the dominant allele. It must also be noted that part of the phenotype observed for these mutants may result from their overexpression. Nonetheless, by comparing their phenotypes with that of wild-type FtsZ overexpression, it is possible to gain some insight into their defects.

Wild-type cells overexpressing mutant 109 had a phenotype similar to cells overexpressing mutant 142; most of the cell area was deeply constricted. However, the unconstricted areas were short and confined to the cell ends. This is probably because, as with mutant 142 expressed in wild-type cells, wild-type FtsZ was available to initiate cell division. Immunolocalization of cells overexpressing mutant 109 showed that FtsZ was localized exclusively in the constricted area, indicating that both the wild-type and the mutant forms of FtsZ were participating in Z ring formation. Like E142, G109 lies in the GTP binding site and contacts the nucleotide. Thus, it is likely that both mutations have altered the GTP-binding or hydrolysis activity of FtsZ.

Mutation 216 is located on the surface that is thought to interact with the GTP-binding surface of the next FtsZ monomer in the protofilaments and participate in GTP hydrolysis. Therefore, mutant 216 may have a defect in FtsZ–FtsZ interaction and in GTP hydrolysis. When mutant 216 was overexpressed, the cells exhibited a smooth filamentous phenotype with tight coils, and FtsZ appeared to form random aggregates in these cells (data not shown). The altered cell shape suggests that expression of mutant 216 perturbs the structure of the cell wall, but in a different manner from mutant 198.

Overproduction of mutant 278 in the presence of wild-type FtsZ conferred a phenotype similar to the phenotype conferred by overexpression of mutant 198, suggesting that they may affect similar aspects of FtsZ function. There is a highly acidic region including helix H10 where mutation 278 is located. Charged surfaces play an important role in protein–protein interactions, and this mutant may be compromised in FtsZ–protein interaction.

In conclusion, we have assembled a set of mutants that will enable us to conduct a detailed analysis of FtsZ function at different stages of cell division. For most of the mutants, it should be possible to make single amino acid mutations with the same phenotype. An analysis of the biochemical defects of these mutants coupled with suppression analysis should prove particularly powerful.

Experimental procedures

Bacterial strains, plasmids, medium and growth conditions

Caulobacter crescentus strains were derivatives of strain NA1000. E. coli DH5α was used for cloning and plasmid purification. E. coli S17-1 was used for the conjugal transfer of plasmids to Caulobacter. Caulobacter was grown at 30°C in peptone–yeast extract (PYE) medium. For Caulobacter, nalidixic acid was used at a concentration of 20 µg ml−1, tetracycline at 1 µg ml−1, kanamycin at 20 µg ml−1, chloramphenicol at 0.5 µg ml−1, xylose at 0.3% and glucose at 0.2%. Plasmids used in this study are described in Table 2.

Table 2. Plasmids used in this study.
PlasmidDescription or constructionSource or reference
pUJ142Derivative of pBBR1MCS with the xylX promoterU. Jenal
pUJ142ftsZ ftsZ transcribed by the xylX promoter Din et al. (1998)
pUJ142-(mutant #)Mutant ftsZ transcribed by the xylX promoterThis study
pHB2.0A 2 kb BamHI–HindIII fragment of ftsZ in pSK+ Quardokus et al. (1996)
plac290ftsZWild-type ftsZ in pRKlac290 Kelly et al. (1998)
plac290-(mutant #)A 0.2 kb BamHI–NcoI fragment containing
the ftsZ promoter from pHB2.0 and a 1.5 kb
NcoI–HindIII fragment containing mutant ftsZ
from pUJ142 clones in pRKlac290
This study
pRKlac290Low-copy-number plasmid, IncP, TetR Gober and Shapiro (1992)
pJM661A 0.38 kb PstI–NcoI fragment containing the xylX promoter
in pSK–. An NcoI site was engineered at the ATG start site
J. Maddock
pJM661NZA 0.48 kb NcoI–BamHI fragment containing the N-terminal
163 amino acids of ftsZ from pHB2.0 in pJM661
This study
pBJMA 0.86 kb HindIII–BamHI fragment with the xylX
promoter and N-terminal 163 amino acids of ftsZ
from pJM661NZ cloned into pBGST18
This study
pBGST18A derivative of pBGS18 that contains a RK2 oriT fragment
into the two DraI sites, does not replicate in Caulobacter
M. R. K. Alley

Construction of a xylose-dependent ftsZ strain and cell synchronization

To construct a strain in which expression of ftsZ is dependent on xylose, we placed the N-terminal portion of ftsZ (ftsZ163ΔC) under the control of the xylose-inducible promoter PxylX (Meisenzahl et al., 1997) in a plasmid that cannot replicate in Caulobacter. This plasmid (pBJM) was then introduced into Caulobacter, followed by selection of the kanamycin resistance marker on the plasmid to select for homologous recombination of the plasmid into the chromosome. There are two sites of homology between pBJM and the Caulobacter chromosome, the ftsZ locus and the xylX locus. If pBJM recombines at the ftsZ locus, the single cross-over event will result in a construct in which ftsZ is under the control of the xylX promoter and ftsZ163ΔC is under the control of the ftsZ promoter (Fig. 1A). We know from previous work that ftsZ163ΔC is a null allele (Din et al., 1998). Thus, if pBJM recombines at the ftsZ locus, cell growth should be xylose dependent. After introduction of pBJM into wild-type NA1000, 15% of the colonies were found to be xylose dependent. Two of these xylose-dependent colonies were picked, and the kanamycin resistance marker of the plasmid was transduced into wild-type NA1000 to get a clean genetic background. One of the transductant strains was used in subsequent experiments and is called YB1585.

For cell cycle studies, the Pxyl–ftsZ strain was grown to late-log phase in PYE medium containing kanamycin and xylose. Swarmer cells were isolated by Ludox density gradient centrifugation (Evinger and Agabian, 1977) and released into PYE medium containing either xylose or glucose at a starting OD of 0.4. Samples were taken at different time points and observed under the microscope. For immunoblotting of swarmer and stalked cells, both swarmer and stalked cell bands were collected after Ludox density gradient centrifugation. The cells were washed, adjusted to the same cell density and processed for immunoblotting.

Caulobacter FtsZ structure modelling

The structure of the N-terminal 330 amino acids of Caulobacter FtsZ (including the NTR region and part of the spacer region) was modelled based on the crystal structure of M. jannaschii FtsZ using insightII software 95.0 (Biosym/MSI). The primary sequences of Caulobacter ftsZ and M. jannaschii ftsZ were aligned. All the secondary structures of M. jannaschii FtsZ, except for H0, were considered to be a structurally conserved region and maintained in the modelled protein. The loop between H6 and H7 (amino acids 176–181), the loop between H9 and S8 (amino acids 254–259), the N-terminus (amino acids 1–14) and the C-terminus (amino acids 319–330) were not conserved between these two proteins and were remodelled using energy minimization (Höltje and Folkers, 1996). All the side-chain positions were adjusted using energy minimization to a convergence of less than 0.1 using an ESFF forcefield.

Site-directed mutagenesis

Site-directed mutagenesis was performed using the QuikChange site-directed mutagenesis kit (Stratagene). Oligonucleotides were designed to contain 9–12 nucleotides of identity on both sides of the mutation site and to create a restriction site. Mutant plasmids were identified by restriction digestion. The entire mutant ftsZ genes were sequenced to verify the nucleotide changes and rule out the possibility of secondary mutations.

General DNA manipulations, sequencing and immunoblotting

General DNA manipulations were performed as described previously (Ausubel et al., 1989; Brun et al., 1994). DNA sequencing of ftsZ mutants was carried out with the ThermoSequenase dye terminator cycle sequencing premix kit (Amersham) using half the recommended reaction volume. pUJftsZ mutant DNA (200–300 ng) was used as template. Five different ftsZ primers were used to determine the full ftsZ sequence in all mutants constructed. The primers used were: ftsZ491, 5′-ATCGGCGAGCACCTCGACGGCG-3′; ftsZ704Rev, 5′-CTGGTTCGGAATGACGATCAGG-3′; ftsZ885, 5′-GATGGGCAAGGCGATGATGGGC-3′; ftsZseq1327, 5′-CGTGGCCTTCGCGCCCGAGCCG-3′; and ftsZ1809Rev, 5′-AACCCCGGCGAGAGTCGCAAG-3′. Primer (5–20 pmol) was used for the reaction. Sequencing gels were run by the Institute for Molecular and Cellular Biology at Indiana University. Immunoblotting was performed essentially as described previously (Quardokus et al., 1996). Anti-FtsZ crude serum was used at a dilution of 1:6000, and the secondary antibody (goat anti-rabbit IgG (H+l)–horseradish peroxidase conjugate (Gibco BRL) was used at 1:15 000.

Immunofluorescence and microscopy

For immunofluorescence microscopy, cells were added directly from cultures to a fixation solution containing (final concentration) 2.5% paraformaldehyde, 30 mM NaPO4, pH 7.5, and incubated at room temperature for 15 min followed by 40 min on ice. Cells were washed three times in 1× PBS. The final pellet was resuspend in 100 µl of GTE (50 mM glucose, 20 mM Tris, pH 7.5, 10 mM EDTA). Lysozyme was added to a final concentration of 2 mg ml−1, and the cells were immediately applied to polylysine-coated slides. After 2 min, the excess cells were aspirated, and the attached cells were allowed to dry completely. Cells were rehydrated with PBS, blocked in PBS containing 2% BSA and incubated in a humid chamber at room temperature for 15 min. Cells were incubated overnight with affinity-purified anti-FtsZ primary antibody diluted (1:100) in PBS containing 2% BSA, washed 10 times with PBS and incubated with goat anti-rabbit fluorescein isothiocyanate (FITC)-conjugated secondary antibody at a dilution of 1:100 for 1 h, and washed 10 times with PBS. Cells were observed with SlowFade (Molecular Probes).

Light photomicroscopy and epifluorescence photomicroscopy were performed on a Nikon Eclipse E800 light microscope with a 100× Plan Apo oil objective. Images were captured with a Princeton Instruments cooled charge-coupled device camera and the metamorph imaging software package V.3. Transmission electron microscopy was performed as described previously (Janakiraman and Brun, 1999) using a Jeol JEM-1010 electron microscope at 60 kV. Cells were stained with 7.5% uranyl magnesium acetate for 5 min.


We thank members of our laboratory for critical reading of the manuscript, and Marty Pagel for help with the molecular modelling. We would like to thank anonymous reviewers for insightful comments and questions about the mutant phenotypes. This work was supported by a National Science Foundation Career Award (MCB-9733958) and by National Institutes of Health grant GM51986 to Y.V.B.