INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

FtsZ plays a key regulatory and structural role during bacterial cell division. It self-assembles into a ring structure on the inner face of the cytoplasmic membrane at the division site and remains at the leading edge of the invaginating septum (6, 28, 33). The FtsZ concentration appears to be an important regulator of its assembly into the ring, because too little FtsZ results in few rings and the formation of nonseptate filaments while a moderate excess of FtsZ results in division at the cell poles in addition to midcell (10, 36). Because the midcell localization of all of the other known essential cell division proteins requires FtsZ, the FtsZ ring is postulated to form the framework that recruits other components of the putative division machinery. FtsZ has been shown to interact directly with FtsA and ZipA (12, 14, 21, 34) and appears to recruit them independently to the FtsZ ring (3, 15, 16, 19). Because many, and perhaps all, of the periplasmic cell division proteins depend upon FtsA for proper localization, and FtsA is well conserved in many bacteria, FtsA may serve as a molecular bridge between the FtsZ ring and the periplasmic proteins.

FtsZ can be thought of as a bacterial cytoskeletal organizer (22). This idea is especially compelling because of the functional and structural homology between FtsZ and tubulin, the key eukaryotic cytoskeletal protein. The crystal structures of FtsZ and tubulin exhibit striking similarities (17). Like tubulin, FtsZ hydrolyzes GTP (11, 25, 31), forms protofilaments in vitro (7, 13) whose assembly depends on GTP (26, 37), and has a similar response to hydrophobic probes (38). The cytoskeletal behavior of FtsZ is also evident in the spiral-shaped septa formed by similarly spiral-shaped FtsZ polymers, indicating that the shape of the FtsZ structure can determine the shape of the resulting septum (2).

Chloroplasts and nearly all prokaryotes have an FtsZ homolog. Based on sequence alignments among the many species from which it has been isolated, FtsZ can be divided into three major domains: a large, conserved N terminus, a variable linker domain, and a small, highly conserved C terminus. The N-terminal region of approximately 320 residues is highly conserved throughout all FtsZs and includes the GTP binding motif that is also found in tubulins. The crystal structure of Methanococcus jannaschii FtsZ shows that residues 38 to 227 in this region form a Rossmann fold-like GTPase domain (17). We previously reported that a truncated Escherichia coli FtsZ containing only this N-terminal domain (amino acids 1 to 316) fused to GFP was able to form large fluorescent polymers in vivo and in vitro (19, 37). This indicated that the C terminus is dispensable for polymerization. In contrast, a deletion of 38 residues at the extreme N terminus abolished the localization of FtsZ-GFP protein to rings in vivo, suggesting that the N terminus of FtsZ is essential for ring assembly (19). Similar conclusions have been drawn from yeast two-hybrid studies of self-association of FtsZ (12, 34).

The linker domain is characterized by the variability of its sequence and its length. While this region is fairly short in most organisms, on the order of 40 to 50 amino acids, several -proteobacterial species contain large linkers of 150 to 200 amino acids that are proline and glutamine rich (23, 29). The specific function of this linker region has not yet been addressed.

Finally, the extreme C-terminal region is variable in length but contains a small consensus sequence of 6 to 8 highly conserved residues that we refer to as the C-terminal core domain (Fig. 1). The core is conserved throughout most bacteria and chloroplasts, but not the archaea. The extreme C terminus of the M. jannaschii FtsZ, therefore, lacks the sequence conservation of the bacterial core domains; in addition, it was disordered in the crystal and hence not resolved (17). As a result, the structure and function of the core has not been studied in detail.

View larger version (43K): [in this window] [in a new window]   FIG. 1.   Alignment of C-terminal core domains of diverse FtsZ homologs; the bottom two are plant chloroplast homologs. The invariant proline residue, corresponding to P375 of E. coli FtsZ, is highlighted in boldface. The underlined residues denote residues changed in this study to an alanine. Arabidopsis thaliana FtsZ refers to one of at least two chloroplast homologs; the other homolog lacks the homology to the core. Archaeal FtsZs also lack this domain, as do those from other Mycoplasma species.

Recently, it was found that a Caulobacter crescentus FtsZ with the C-terminal 24 residues, including the core domain, deleted displayed a dominant-negative phenotype and was unable to interact with FtsA in a yeast two-hybrid assay (12). Although this study did not specifically pinpoint the core residues, it suggested that this domain is important for the FtsZ-FtsA interaction. In addition, studies with an FtsZ with the entire C-terminal domain (linker plus core) deleted indicated that either the linker domain, the C-terminal core, or both are required for recruitment of FtsA and/or ZipA (12, 16). However, the core domain was not specifically targeted in these studies.

In this paper, we characterize the functional domains of the C terminus of FtsZ by constructing a series of C-terminal truncations and point mutations in E. coli ftsZ. We then test the abilities of these mutant proteins to localize and interact with FtsA and ZipA. Our data suggest that the extreme C terminus of FtsZ is not required for its localization but may be important for its interaction with FtsA and ZipA. Alteration of several consensus residues in the core region appears to affect only the interaction with FtsA. Furthermore, several of these point mutant FtsZs display dominant-negative phenotypes.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. E. coli strains and plasmids are listed in Table 1. Strain JM105, containing lacI^q on an F' plasmid, was used as the wild-type E. coli host strain. JFL101 (obtained from J. Lutkenhaus, University of Kansas Medical Center), containing the ftsZ84(Ts) allele, was used as the thermosensitive ftsZ mutant strain. The ftsZ depletion strain, here designated WM746, is WX7/pCX41 (35) and was obtained from W. Cook and L. Rothfield, University of Connecticut Health Center. The standard growth medium was Luria-Bertani (LB) medium, containing 0.5% NaCl, or LBNS, which contains no added NaCl. This medium was supplemented with ampicillin (50 µg per ml), chloramphenicol (20 µg per ml), or tetracycline (10 µg per ml) when needed. E. coli wild-type strains were grown at 37°C, while temperature-sensitive strains were grown at 32°C (permissive) or 42 to 44°C (nonpermissive) on LBNS medium containing appropriate antibiotics.

Plasmid construction. The C-terminal truncation mutants of ftsZ were generated by PCR amplification. A single forward primer (5'AAGAGCTCGGAGAGAAACTATG) was used, which includes the ftsZ native ribosome binding site as well as a SacI site at the 5' end. The reverse primers were designed to be complementary to the sequences encoding different regions of the FtsZ C terminus and also included a stop codon as well as an XbaI site at the 5' end. Primers for the FtsZ-GFP fusion proteins, corresponding to these C-terminally truncated FtsZ mutants, did not contain a stop codon and were in frame with the sequence encoding the N terminus of GFP. The sequences of the reverse primers for generating these FtsZ C-terminal deletion mutants were as follows: 5'AATCTAGATTAATAATCCGGCTCTTTCG (ZC1), 5'TTTTCTAGATTAGTCATTCACGACTTTAG (ZC1.1), 5'AAATCTAGATTACGGAGCCATCCCATGC (ZC1.2), 5'AAATCTAGATTAAACCTGCTTATTGGTC (ZC1.3), and 5'AATCTAGATTAGCCGATACCTGTCGCAAC (ZC2). The sequences of the reverse primers for constructing the corresponding FtsZ-GFP fusion proteins were 5'AATCTAGAATAATCCGGCTCTTTCG (ZC1), 5'AAATCTAGAGCCGATACCTGTCGCAAC (ZC2), 5'AAATCTAGAGATAACCACAGTCGCG (ZC3), and 5'AAATCTAGACTCATCCAGACGCAGG (ZC4).

Complementation tests. To test the function of the FtsZ mutant alleles, we transformed the versions cloned in pWM176 into JFL101 and WM1099, an ftsZ84(Ts) mutant and an ftsZ depletion strain, respectively. Because pWM176 derivatives and the original ftsZ depletion strain, WM746, were both tetracycline resistant (Tet^r), WM746 was made Tet^s by replacing the linked Tet^r marker in leu with a Kan^r marker from CAG12131 by P1 transduction; the resulting strain became WM1099.

Immunoblot analysis. JFL101 cells containing the ftsZ derivatives in pWM176 were grown at 30°C to exponential phase and then induced with 10 µM IPTG for 90 min at 42°C. JM105 cells harboring the same plasmids were induced with 5 µM IPTG for 3 h at 37°C. The cells were then collected, and total protein was quantitated with the bicinchoninic acid protein assay (Pierce). A total of 25 µg of protein for each sample was separated on sodium dodecyl sulfate-12% polyacrylamide gels and transferred to nitrocellulose membranes (0.45-µm pore size; Micron Separations Inc.). The amplified alkaline phosphatase Immuno-Blot Kit (Bio-Rad) was then used for immunoblotting. Purified anti-FtsZ antiserum at 1:200 dilution was used as the primary antibody, and biotinylated goat anti-rabbit immunoglobulin G at 1:1,000 dilution was used as the secondary antibody.

Yeast two-hybrid analysis. The Saccharomyces cerevisiae host strain L40 carries a LexA binding DNA sequence fused with lacZ on the chromosome, so the expression of lacZ can be activated upon binding of LexA (5). The plasmids used were pACT2.2, containing the GAL4 activation domain (obtained from S. Elledge, Baylor College of Medicine), and pLEXA, containing the LexA DNA binding domain (obtained from J. Jones, M. D. Anderson Cancer Center). To clone ftsA into pLEXA, ftsA was amplified by PCR, using 5'AAACATATGATCAAGGCGACGGAC as the upstream primer and 5'TTTGGATCCTTAAAACTCTTTTCGC as the downstream primer. The PCR product was digested with NdeI, followed by fill in of the overhangs with Klenow fragment and digestion with BamHI. The cleaved product was ligated to SmaI/BamHI-cleaved pLEXA, and the resulting plasmid was designated pWM1208. For cloning ftsZ into pACT2.2, ftsZ was amplified by PCR with 5'AACATATGTTGAACCAATGGAAC as the upstream primer and 5'AAGGATCCATCAGCTTGCTTACGC as the downstream primer. The PCR fragment was cleaved with NdeI and BamHI and cloned into the NdeI and BamHI sites of pBCIISK(+). The C-terminal ftsZ mutant alleles were cloned into the same vector by replacing the C-terminal coding region of ftsZ in pBCII-ftsZ between the SacII site within ftsZ and the BamHI site. The ftsZ gene and its mutant derivatives were subcloned between the NdeI and BamHI sites on pACT2.2. These two plasmids (pWM1208 and pACT2.2-ftsZ or its derivatives) were then transformed into the yeast reporter strain L40. -Galactosidase activities were estimated by using a filter assay, with X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) as a substrate (4).

Microscopic techniques. Immunofluorescence staining in combination with phase-contrast imaging and nucleoid staining with 4',6-diamidino-2-phenylindole (DAPI) were carried out as described previously (39) with minor modifications. The concentration of lysozyme used to permeabilize the cells was varied from 0.8 to 2 mg per ml for different strains to obtain the best results. For FtsZ staining, we used affinity-purified anti-FtsZ at a final dilution of 1:20. To stain FtsA-GFP or ZipA-GFP, polyclonal anti-GFP antiserum was used at a 1:200 dilution. To generate the anti-GFP, we purified GFP from a strain carrying the mut2GFP-overproducing plasmid (9). The antigen was sent to Cocalico Biologicals Inc. (Reamstown, Pa.) for antibody production in rabbits. Anti-FtsZ was similarly obtained from rabbits by using purified overproduced E. coli FtsZ as an antigen and was subsequently affinity purified. All microscopic images were taken on an Olympus BX60 microscope, captured in RGB mode with an Optronics DEI-750 camera and a Scion CG-7 frame grabber, and manipulated in Adobe Photoshop. The images were changed to grayscale mode for publication.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Function of C-terminal deletions of FtsZ. To identify the C-terminal domains of E. coli FtsZ required for in vivo function, we constructed a series of C-terminal deletions of E. coli ftsZ (Fig. 2) and placed them under the control of the IPTG-inducible tac promoter in the IncP plasmid derivative pWM176 (24). We first examined whether the mutant ftsZs were capable of complementing the ftsZ84 temperature-sensitive strain JFL101, which is nonviable at 42°C. A strain containing pMK4 (pWM176 carrying the wild-type ftsZ) was unable to form colonies on LBNS plates at 42°C in the absence of IPTG, indicating that uninduced expression of ftsZ from the plasmid was insufficient to support cell division. Therefore, different IPTG concentrations were used to induce complementing levels of FtsZ. In a range of 10 to 40 µM IPTG, strains harboring pMK4 but not pWM176 formed fast-growing colonies with normal plating efficiency at 42°C, confirming complementation of the ftsZ84 mutant by the wild-type ftsZ on pMK4. When identical induction conditions were used with the deletion derivatives, none of these, including an N-terminal-deletion control, were able to complement ftsZ84 (Table 2).

View larger version (26K): [in this window] [in a new window]   FIG. 2.   FtsZ truncation and point mutant derivatives. ZC* represents six mutants with single-residue changes in the C-terminal core. Slanted hatches represent the N-terminal conserved domain, horizontal hatches represent the C-terminal core, and the checkered pattern indicates the C-terminal core that contains mutations. The numbers indicate the beginning or ending residues relative to those of wild-type FtsZ. ZC2, ZC3, ZC4, and ZN all contain C-terminal GFP fusions (see Materials and Methods).

Function of point mutations in the C-terminal core of FtsZ. The failure of ZC1 (amino acids 1 to 371; pWM737), which lacks only the C-terminal 12 residues, to complement the ftsZ84(Ts) or the ftsZ null mutant suggested that residues 372 to 383 are essential for the full function of FtsZ. Sequence alignments of FtsZs from different species revealed that this extreme C terminus is quite highly conserved, with at least 6 residues present in most FtsZs and a central proline residue, corresponding to P375 of E. coli FtsZ, that is particularly well conserved (Fig. 1). To identify the critical amino acids in this region, we constructed six ftsZ mutant alleles with each conserved residue individually changed to alanine (Fig. 2). Five of these point mutants (ZL372A, ZI374A, ZP375A, ZF377A, and ZL378A) failed to complement either the ftsZ84(Ts) or the ftsZ null mutant (Table 2). Three mutant FtsZ derivatives (ZL372A, ZI374A, and ZP375A) exhibited a dominant-negative phenotype similar to that of ZC1.1 (Table 2).

View larger version (18K): [in this window] [in a new window]   FIG. 3.   Western blot analysis of FtsZ and its derivatives expressed in JM105 (wild type) and JFL101 (ftsZ84). (A) JM105 derivatives were grown exponentially at 37°C and then induced with 5 µM IPTG for 180 min. (B) JFL101 derivatives were grown at 42°C for 30 min before being induced with 10 µM IPTG for 90 min. The cells were then collected, and total protein was quantitated by bicinchoninic protein assay (Pierce). An equivalent amount of protein (25 µg) was loaded onto each lane, followed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and immunoblotting. Lane 1, cells containing pWM176; lane 2, pMK4 (wild-type ftsZ); lane 3, pWM1202 (ZL372A); lane 4, pWM738 (ZD373A); lane 5, pWM1203 (ZI374A); lane 6, pWM739 (ZP375A); lane 7, pWM1204 (ZF377A); lane 8, pWM1205 (ZL378A); lane 9, pWM737 (ZC1); lane 10, pWM930 (ZC1.1); lane 11, pWM931 (ZC1.2); lane 12, pWM932 (ZC1.3); lane 13, pWM1201 (ZC2).

Localization patterns of FtsZ mutant derivatives. When synthesized from plasmids, GFP-tagged FtsZ derivatives lacking the N-terminal 38 residues (ZN, containing amino acids 38 to 383) (19), the C-terminal 89 residues (ZC3 with amino acids 1 to 294), or the C-terminal 119 residues (ZC4 with amino acids 1 to 274) did not exhibit detectable fluorescent foci or rings in cells and instead displayed unlocalized fluorescence (data not shown). These observations, summarized in Table 3, suggest that residues 1 to 316 may be required for the proper localization of FtsZ to division sites.

View larger version (64K): [in this window] [in a new window]   FIG. 4.   Localization of FtsZ and its derivatives in JFL101 [ftsZ84(Ts)] cells. JFL101 derivatives containing various plasmids were grown at 42°C for 30 min, induced with 10 µM IPTG for 60 min, and then fixed for staining. Simultaneously, FtsZ was stained by anti-FtsZ and visualized by IFM (A to D), nucleoids were stained with DAPI and visualized by fluorescence microscopy (A' to D'), and cell morphologies were visualized by phase-contrast microscopy (A" to D"). (A to A") JFL101 containing pWM176; (B to B") JFL101 containing pMK4 (wild-type ftsZ); (C to C") JFL101 containing pWM1202 (ZL372A); (D to D") JFL101 containing pWM932 (ZC1.3).

View larger version (57K): [in this window] [in a new window]   FIG. 5.   Localization of FtsZ and its derivatives in WM1099 (ftsZ depletion strain) cells. WM1099 derivatives were grown at 42°C for 5 h before induction with 5 µM IPTG for 40 min and fixation. Simultaneously, FtsZ was stained by anti-FtsZ and visualized by IFM (A to D), nucleoids were stained with DAPI and visualized by fluorescence microscopy (A' to D'), and cell morphologies were visualized by phase-contrast microscopy (A" to D"). (A to A") WM1099 containing pWM176; (B to B") WM1099 with pMK4 (wild-type ftsZ); (C to C") WM1099 with pWM1202 (ZL372A); (D to D") WM1099 with pWM932 (ZC1.3).

Protein-protein interactions between FtsA and FtsZ deletion derivatives. To examine interactions between FtsA and FtsZ mutant derivatives, we developed a simple and novel in vivo protein-protein interaction assay. This assay was based on our previous observations that when FtsA-GFP is co-overproduced with FtsZ, fluorescent spiral structures assemble throughout the cell. Similar spirals are often found after overproduction of FtsZ-GFP, but not after overproduction of FtsA-GFP. Because FtsZ-GFP, but not FtsA-GFP, was able to form extensive spirals when overexpressed, we previously postulated that fluorescent spirals in cells overproducing FtsA-GFP and FtsZ resulted from association of FtsA-GFP with FtsZ polymers (24). This assay was then used to test interspecies FtsZ-FtsA interactions in situ and revealed a significantly stronger interaction between rhizobial FtsA-GFP and its cognate FtsZ than with E. coli FtsZ (21).

View larger version (52K): [in this window] [in a new window]   FIG. 6.   Association of FtsA-GFP with FtsZ spirals in JFL101 [ftsZ84(Ts)]. JFL101 cells containing plasmids synthesizing FtsA-GFP and an FtsZ derivative were grown at 42°C for 30 min and then induced with 200 µM IPTG for 60 min. Parallel samples were fixed and stained with anti-GFP antibodies for FtsA-GFP and stained with anti-FtsZ antibodies for FtsZ, followed by IFM. (A to E) FtsA-GFP; (A' to E') FtsZ. (A and A') WM1234 (JFL101 containing pWM633 [ftsA-GFP] and pWM176); (B and B') WM1235 (pMK4 [wild-type ftsZ]); (C and C') WM1242 (pWM738 [ZD373A]); (D and D') WM1244 (pWM739 [ZP375A]); (E and E') WM1236 (pWM737 [ZC1]).

Protein-protein interactions between FtsA and FtsZ point mutant derivatives. This possibility prompted us to test the interactions between FtsA-GFP and the C-terminal core point mutants of FtsZ in the same assay. When stained with anti-FtsZ, cells overproducing wild-type FtsZ (Fig. 6B') or any of the FtsZ point mutants (Fig. 6C' and D' and data not shown), except the unstable mutants ZI374A and ZF377A, exhibited fluorescent spirals. This was expected, because the same mutants were all able to form fluorescent rings in ftsZ84 cells. However, anti-GFP staining revealed fluorescent FtsA-GFP spirals when FtsA-GFP was cosynthesized with ZD373A and ZL378A (Fig. 6C, showing ZD373A, and data not shown) but diffuse fluorescence with weak background patterns when cosynthesized with ZL372A and ZP375A (Fig. 6D, showing ZP375A, and data not shown). These results suggest that L372A and P375A, the two stable point mutants within the conserved core that cannot complement an ftsZ mutant, are also defective in efficiently interacting with FtsA-GFP.

Yeast two-hybrid analysis of FtsA interactions with mutant FtsZs. To test independently whether there was indeed a reduced interaction between FtsA-GFP and the truncated FtsZs, ZL372A, or ZP375A, as observed in the JFL101 (ftsZ84) strain, we used a yeast two-hybrid assay to measure FtsZ-FtsA interaction. We cloned native ftsA into the yeast vector pLEXA that carries the LexA DNA binding domain. Either wild-type ftsZ or its mutant alleles were fused in frame to the coding region of the GAL4 activation domain (GAL4AD) in pACT2.2. Combinations of the hybrid plasmids were then introduced into the yeast reporter strain L40, and the resulting cotransformants were tested for -galactosidase activity. The pLEXA-ftsZ construct could not be used because of high background activity (Table 4). However, control experiments showed that cells containing pLEXA-ftsA and pACT2.2-virE2 (VirE2 is an unrelated protein that is part of the Agrobacterium tumefaciens T-DNA transport system) did not have background -galactosidase activity (Table 4). Cotransformants expressing both LexA-FtsA and GAL4AD-FtsZ displayed considerable -galactosidase activity: the colonies were blue within 2 h of incubation on a filter containing X-Gal. On the other hand, when LexA-FtsA was combined with a GAL4AD fusion of any mutant FtsZ protein except ZD373A, the yeast colonies were white on the filters after overnight incubation, suggesting a lack of interaction with FtsA and the FtsZ mutant. A positive but somewhat weaker interaction (pale blue on the filter) was detected between FtsA and ZD373A (Table 4). Taken together, the results from the yeast two-hybrid assay were consistent with the E. coli in vivo interaction assay in the thermosensitive ftsZ84 mutant. The data are summarized in Table 5.

Interaction between ZipA and FtsZ mutant derivatives. The interaction between ZipA and FtsZ was initially demonstrated by affinity blotting (14) and recently demonstrated by yeast two-hybrid analysis and cosedimentation (16, 30). We wanted to test whether an interaction could be detected between ZipA-GFP and FtsZ in our E. coli in vivo assay. As with the FtsA-GFP studies, a zipA-GFP fusion was cloned downstream of either the lac promoter in pBC or the BADp promoter in pBAD30 and was introduced along with pWM176 carrying ftsZ or its derivatives into either the ftsZ84 thermosensitive strain JFL101 or the ftsZ depletion strain WM1099. Cells harboring both plasmids were grown under conditions of FtsZ inactivation or depletion, as described above for FtsZ-FtsA interactions, and ZipA-GFP and FtsZ were detected in cells by IFM with anti-GFP and anti-FtsZ antibodies, respectively.

View larger version (61K): [in this window] [in a new window]   FIG. 7.   Binding of ZipA-GFP to FtsZ rings and spirals in the ftsZ depletion strain WM1099. WM1099 containing pWM1206 (zipA-GFP) and various ftsZ derivatives were grown at 42°C for 4 h and then induced by 0.2% L-arabinose and 10 µM IPTG for 60 min. As described in the legend to Fig. 6, parallel samples were taken. One was stained with anti-GFP for ZipA-GFP and the other was stained with anti-FtsZ for FtsZ. The cells were observed by IFM for ZipA-GFP (A to D) and for FtsZ (A' to D'). (A and A') WM1221 (WM1099 containing pWM1206 [zipA-GFP] and pWM176); (B and B') WM1222 (pWM1206 plus pMK4 [wild-type ftsZ]); (C and C') WM1231 (pWM1206 plus pWM739 [ZP375A]); (D and D') WM1223 (pWM1206 plus pWM737 [ZC1]).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have shown that the C-terminal conserved core domain of FtsZ is essential for function, because a plasmid expressing a 12-amino-acid deletion of this domain cannot complement a chromosomal ftsZ mutation. Additional evidence for the importance of this conserved region comes from alanine scan mutagenesis. Most of the single-residue alterations between L372 and L378, with the exception of ZD373A, cannot fully substitute for the native protein. Even cells synthesizing ZD373A as the only functional FtsZ exhibited moderate filamentation, indicating that despite its ability to allow an ftsZ mutant to form colonies at the restrictive temperature, this mutant FtsZ is not completely normal. Synthesis of wild-type FtsZ in the same plasmid system in the same strains, on the other hand, resulted in mostly normal-length cells.

The loss of the entire C-terminal core had a more severe effect on FtsZ function than any one of the single-amino-acid substitutions. For example, after depletion of FtsZ in WM1099, the stable point mutants of FtsZ (ZL372A, ZD373A, ZP375A, and ZL378A) assembled into rings at potential division sites within the filaments. However, the C-terminal-truncation mutants containing at least residues 1 to 316, such as ZC1, exhibited a regular punctate pattern within the filaments. This result indicates that they can localize to potential division sites and aggregate but cannot assemble into normal rings like the point mutants. The inability of these deleted derivatives to form normal rings was somewhat unexpected because ZC2, containing only residues 1 to 316, is still able to form large polymers in vitro when fused to GFP (37).

Despite their failure to form rings in the depletion strain, these C-terminally truncated FtsZs with at least residues 1 to 316 formed rings in the ftsZ84(Ts) mutant JFL101 at 42°C. This strain-specific difference in the behavior of the truncated FtsZs can be rationalized by partial function of the thermoinactivated FtsZ84. In JFL101, several thousand FtsZ84 proteins are estimated to be present per cell at 42°C, despite the fact that rings are not formed under these conditions. The ability to suppress this defect by overproducing FtsZ84 (27) or increasing ZipA levels (30) indicates that FtsZ84 is competent for assembly into functional rings but is less efficient at higher temperatures and requires a higher critical concentration than the wild-type protein. In our experiments, we speculate that exogenous synthesis of one of our mutant FtsZs in JFL101 at 42°C increases the total FtsZ concentration, stimulating the assembly of FtsZ84, which in turn allows the incorporation of the truncated FtsZs into a morphologically normal ring.

The ability of the stable point mutants and some of the truncated FtsZs to form rings on their own or in concert with otherwise thermoinactivated FtsZ84 indicates that the structures of these proteins are not significantly perturbed by the mutations. Moreover, dominant-negative effects were displayed by several of the mutant proteins even when present at less than wild-type levels. Such effects indicate that these mutant proteins retain normal structural motifs that allow them to interfere with the function of the native FtsZ. The most likely avenue for such interference is by coassembling with native FtsZ to form mixed rings. Why would such rings, which appear normal by fluorescence microscopy, be defective for septation? One hypothesis is that the heteropolymer cannot properly interact with a downstream protein such as FtsA. Although some of the point mutants that exhibit defects in associating with FtsA in our assays (ZL372A, ZI374A, and ZP375A) exhibit a clear dominant-negative phenotype, other mutants, such as several of the truncation mutants that are also defective in FtsA interactions, are not dominant negative. Therefore, this explanation for the phenotypes is unlikely to be correct. A defect in interacting with ZipA would also not explain the dominant-negative phenotypes because there is little correlation between these two characteristics. It is also intriguing that the smallest deletion (ZC1) has no dominant-negative phenotype but a larger deletion (ZC1.1; deletion of 23 amino acids) does and still-larger deletions (ZC1.2 and 1.3 and ZC2) do not. By inference, the dominant-negative phenotype of ZI374A may be caused by the posttranslational truncation of the protein observed on immunoblots. Although the truncation point for ZI374A is not yet established, we speculate that the resulting protein is similar in structure and function to the ZC1.1 truncation.

Interestingly, a 24-amino-acid C-terminal truncation of C. crescentus FtsZ exhibited a dominant-negative effect (12) similar to our 23-amino-acid truncation of E. coli FtsZ (ZC1.1), as well as some of our point mutants. As with our truncated E. coli FtsZs synthesized in cells with native FtsZ depleted, the truncated C. crescentus FtsZ localized in a punctate pattern in C. crescentus cells. Moreover, truncated C. crescentus FtsZ was also defective in interacting with C. crescentus FtsA in the yeast two-hybrid system. The similarity of these results suggests that the dominant-negative effects may have general significance. However, for the reasons mentioned above, it is unlikely that the defect in FtsZ-FtsA interactions is the cause of the dominant-negative phenotype.

The results described here suggest that the last 12 amino acids of FtsZ are important for its efficient interaction with both ZipA and FtsA. This finding is consistent with the determinants of C. crescentus FtsZ that are needed for FtsA recruitment (12) and also with recent results implicating the C terminus of FtsZ in ZipA recruitment (16). Our studies of single-residue mutants in the C-terminal core have pinpointed two conserved residues, L372 and the invariant residue P375, that appear to have important roles in FtsA recruitment. ZipA recruitment, on the other hand, does not appear to be affected by these point mutations, suggesting that other single-residue changes may be necessary to detectably affect ZipA binding.

Our results demonstrate a correlation between noncomplementation by the FtsZ mutants and defective interaction with FtsA. However, it is not yet clear from our data or various genome sequence comparisons whether the defect in FtsA interaction is the sole reason for the noncomplementation. Mycobacterium tuberculosis, for example, lacks ftsA yet has the equivalent of L372 and P375; conversely, Aquifex aeolicus has ftsA but has only the P375 and not the L372 equivalent (Fig. 1). Our working hypothesis is that at least our C-terminal point mutants may be defective for complementation solely because they are defective for FtsA interaction. This idea seems reasonable, because these mutant proteins can (i) form rings at division sites by themselves and (ii) interact with ZipA.

The effects of ZL372A and ZP375A on the production of fluorescent spirals with FtsA-GFP were clear and reproducible with the JFL101 [ftsZ84(Ts)] strain. However, the results with the WM1099 depletion strain were less conclusive, as spirals were sometimes observed with all the stable point mutant FtsZs. One likely explanation for this discrepancy is that because FtsA-GFP was synthesized in the depletion strain with the strong araBAD promoter, FtsA-GFP levels were about 10- to 20-fold higher than those of FtsZ in the experiments with the depletion strain than with JFL101. This high level of protein may sometimes have forced an interaction between FtsA-GFP and ZL372A or ZP375A proteins that does not normally occur efficiently at lower FtsA-GFP levels. Because FtsA is normally present at 50- to 100-fold-lower levels than FtsZ, we favor the idea that the results in the JFL101 background are more physiologically relevant and that ZL372A and ZP375A are truly defective in binding to FtsA. Importantly, the yeast two-hybrid data are entirely consistent with this conclusion. Future experiments with purified proteins should allow more precise biochemical definition of the binding affinities between the various FtsZ mutants and FtsA. Such experiments should also be able to determine if the mutations in the C-terminal core disrupt an actual binding surface for FtsA and/or ZipA or instead indirectly cause structural changes in FtsZ that weaken the binding of these proteins.

Why does synthesis of high levels of FtsZ in conjunction with FtsA-GFP or ZipA-GFP result in FtsZ spirals? FtsZ spirals caused by the ftsZ26 mutation can function normally in cell division (2), although spirals caused by overproduction of FtsZ alone (20), FtsZ-GFP, or co-overproduction of FtsZ and FtsA-GFP appear to be nonfunctional (19). The conditions controlling the assembly of spirals and their different morphologies and pitches in the cell are not known. We favor the idea that spirals represent a hyperstable form of FtsZ polymers and that ZipA and FtsA, at least when present at certain levels, are able to promote stabilization of FtsZ polymers. Biochemical and genetic evidence for such a stabilization function for ZipA has recently been reported (30). A better understanding of this phenomenon should prove useful in the design of drugs that, like taxol with microtubules, can inactivate the bacterial division system by hyperstabilizing FtsZ polymers.