![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Previous Article | Next Article 
Journal of Bacteriology, April 2005, p. 2727-2736, Vol. 187, No. 8
0021-9193/05/$08.00+0 doi:10.1128/JB.187.8.2727-2736.2005
Copyright © 2005, American Society for Microbiology. All Rights Reserved.
Jesse Stricker,
Gina Briscoe, and
Harold P. Erickson*
Department of Cell Biology, Duke University, Durham, North Carolina
Received 18 October 2004/ Accepted 7 January 2005
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The tubulin protofilament is a linear strand of subunits connected by longitudinal bonds. Each longitudinal bond is an interface of the plus end of one subunit and the minus end of the subunit above it. The GTP bound to the plus end is sandwiched in this interface. Amino acids that make contact across the interface have been identified for 
- and ß-tubulin (23). Many of the contacting amino acids are conserved between tubulin and FtsZ, including the synergy or T7 loop on the minus end (NxDFxD in most FtsZ), which contacts the GTP of the subunit below it. A number of amino acids on the plus end are also conserved between tubulin and FtsZ (see Table 1, below, for the correspondence of amino acids in FtsZ and tubulin). Other FtsZ residues likely to be in the interface can be identified by their proximity to the GTP on the plus end or to the synergy loop on the minus end.
|
Mutations of amino acids in the protofilament bond interface could provide new insight into protofilament assembly and dynamics. We were especially interested in looking for mutants that had dominant negative effects. We expected that some of these might poison Z-ring assembly substoichiometrically, and this might give an indication of the size of the subassemblies in the Z-ring. For this we needed to develop a system for varying and quantitating the expression level of the mutated FtsZ relative to that of the endogenous wild-type FtsZ.
In the present study we prepared 11 mutants of the bottom interface and 13 mutants of the top interface and characterized them in vivo for complementation and dominant negative effects. We have determined the GTPase activity of each mutant and have examined the GTPase activity of selected pairs of bottom and top mutants. The results provide several novel insights into the interface of the protofilament bond and several surprises that will motivate new experiments and models for the function of the Z-ring.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
500 bp included in the sequence run. Only one inadvertent mutation was discovered in more than 100 sequencing runs. ftsZ genes from Synechocystis, Bacillus subtilis, and Mycoplasma pneumoniae were amplified by PCR from available plasmids and cloned into pJSB2.
Complementation and dominant negative assays.
Complementation tests were performed on plates and in liquid culture as described by Stricker and Erickson (36). Arabinose at 0.05% was used for induction, as expression at this level is near maximum for this construct (5 times the level of endogenous FtsZ) (see Fig. 2, below). Dominant negative tests were modified by the use of BW27783 as host strain, whose utility is explained in Results (14). BW27783 was a gift of Jay D. Keasling (University of California, Berkeley).
|
0.5 to 0.7, harvested, and prepared for quantitative Western blotting as described by Lu et al. (17). Western blotting showed two closely spaced bands: a lower-molecular-weight band that comigrated with purified FtsZ protein and a larger band that corresponded to the FtsZ-myc protein. These bands were quantitated as described by Lu et al. (17) except that goat anti-rabbit immunoglobulin G conjugated to Alexa 488 (Molecular Probes, Eugene, Oreg.) was used as a secondary antibody, and a Molecular Dynamics Storm 860 fluorimager was used to collect data. Immunofluorescence microscopy. JKD7-1/pKD3 cells transformed with pJSB2-derived plasmids expressing mutant ftsZ alleles were grown as overnight cultures in Luria-Bertani (LB) plus 34 µg of chloramphenicol/ml, 100 µg of ampicillin/ml, and 0.2% (wt/vol) glucose at 30°C. Overnight cultures were diluted 1:2,000 in LB plus chloramphenicol and 0.05% (wt/vol) arabinose and grown at 42°C (complementation conditions) until the OD600 reached 0.5 to 0.7, at 240 to 300 min. This incubation time was sufficient for noncomplementing strains to stop growing due to loss of pKD3. Cultures were fixed, permeabilized, and stained as described previously (36). Images were captured using a Zeiss Axiophot microscope with 40x and 100x Plan-Neofluar objectives (numerical aperture, 1.4) and a Photometrics CoolSnap HQ cooled charge-coupled device camera and were processed for optimal contrast with Adobe Photoshop.
Protein purification.
Wild-type and mutant FtsZ proteins were purified using the protocol of Lu et al. (17), with some modifications. For some preparations we omitted Mg and DNase, instead using sonication to reduce the viscosity from DNA. For wild-type FtsZ, we continued to use the 20% ammonium sulfate cut followed by a 25% precipitation, but for most of the mutants we used a single precipitation at 25 to 35% saturated ammonium sulfate. We have added a step of column purification on an anion exchanger. The 25% ammonium sulfate pellet was resuspended in 10 ml of Z buffer (21) (50 mM Tris [pH 7.9], 1 mM EDTA, 10% [vol/vol] glycerol) plus 50 mM KCl. The resuspended pellet was prepared for chromatography by dialysis versus Z buffer plus 50 mM KCl and filtration through 0.2-µm filters. This filtered material was applied to a Resource Q 10/10 column (Amersham Biosciences, Piscataway, N.J.) and eluted with a linear gradient of 100 to 500 mM KCl in Z buffer. This chromatography was done at 4°C. The eluted protein was located by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and peak fractions were pooled. GTP was added to 100 µM, and the pools were aliquoted and stored at –80°C.
Assays for GTP binding and GTP hydrolysis. To assay GTP binding, 0.2 mM GTP was added to 2.5 ml of FtsZ at 20 to 30 µM in HMK buffer (50 mM HEPES [pH 7.7], 5 mM Mg-acetate [MgAc], 350 mM KAc). The protein was loaded on a PD10 gel filtration column (Amersham) equilibrated in HMK buffer. After the sample was loaded onto the column, it was eluted with 3.5 ml of HMK. The first 1.0 ml was discarded, and the next 2.2 ml was collected in a single fraction. The protein concentration was determined by the bicinchoninic acid assay (Pierce, Rockford, Ill.). Bovine serum albumin was used as a standard, and FtsZ concentrations were adjusted to reflect the 0.75 color ratio of FtsZ and bovine serum albumin (18). Perchloric acid was added to 2.5% to the remainder of the sample, and after 30 min on ice the denatured protein was removed by centrifugation. The GTP concentration was determined by UV spectroscopy using an extinction coefficient of 12,700 at 256 nm.
GTPase activity was determined by measuring the concentration of evolved phosphate using an improved sensitivity malachite green assay (11). Proteins were diluted to the desired concentrations (ranging from 0.2 to 40 µM) in HMK buffer. The reaction was initiated by adding GTP to 1 mM at 37°C. Aliquots were taken at 0, 1, 2, 5, 10, and 20 min, transferred to tubes containing 2.3 volumes water and 3.3 volumes of 0.6 M perchloric acid, and placed on ice. After all time points, samples and monobasic potassium phosphate standards were aliquoted in triplicate to microtiter plates. An equal volume of a solution containing 0.78 mM malachite green, 8.3 mM sodium molybdate, 0.05% (vol/vol) Triton X-100, and 0.7 M HCl was added, the plates were incubated at room temperature for 30 min, and the A655 was measured using a Bio-Rad model 550 microplate reader. Free phosphate was plotted as a function of time, and the curves were examined to find a linear region (usually from 2 to 10 min or 5 to 20 min). The slope of the linear region was used to determine the hydrolysis rate. These rates were then plotted versus protein concentration to determine a critical concentration (Cc) and true hydrolysis rate above the Cc for each mutant.
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
-tubulin (where it is involved in GTP hydrolysis of the underlying beta tubulin), and is K and G in ß- and 
-tubulin, where it apparently does not participate in GTP hydrolysis. These synergy loop residues were previously mutated to Cys by Scheffers et al. (34), who kindly provided us with their expression plasmids. These Cys mutants are all blocked for GTPase, but some of them can assemble into protofilaments and/or coassemble with wild-type FtsZ (34) (our observations and data not shown). We therefore made more drastic mutations. D209 was replaced with K, changing the charge and increasing the length of the side chain. F210, which lies beside the synergy loop and appears to support it (Fig. 1), was changed to A. We expected that this might cause the synergy loop to collapse. We also made mutations farther away from the synergy loop. In one case, we made a triple mutant, replacing R271, L272, and D273 with A, K, and K. These three amino acids are not highly conserved in FtsZ but were selected because they are homologous to SMK324-326 of ß-tubulin. Nogales et al. (23) identified these residues as making a longitudinal contact. Finally, we included the mutant D96A in this analysis. We had thought that this might be in the longitudinal interface (Fig. 1), but the recent crystal structure of the FtsZ dimer (25) shows that this is actually about 10 Å above the interface, on a lateral surface.
We used two strategies to select top mutants. We mutated residues that make contact with the GTP molecule, with the goal of blocking GTP binding, and we mutated other nearby residues, with the goal of destabilizing the interface. We added or changed charge in some cases and added or removed bulky hydrophobic groups in other cases.
Test for complementation. We used a complementation system that employs an ftsZ null host strain (JKD7-1) (7) carrying two plasmids. The rescue plasmid pKD3 expresses wild-type FtsZ but is temperature sensitive for replication and is lost at 42°C (7). The tester plasmid (pJSB2) expresses FtsZ (with the mutation of interest) under the control of the araBAD promoter (36). Each strain, carrying a different mutated ftsZ, was first grown overnight in liquid culture at 30°C and then plated on LB agarose containing different concentrations of arabinose. As in our previous study (36), complementing mutations and wild-type ftsZ gave similar response curves. Colonies began to appear at 0.005% arabinose, and their number plateaued at or above 0.05%. (Note that these arabinose levels are much higher than would be expected from Fig. 2, below. However, the experiment in Fig. 2 used the cell line BW27783, and the complementation cell line was JKD7-1. It is likely that JKD7-1 requires higher arabinose levels for induction of FtsZ at the wild-type level.) All of the mutants we tested either fully complemented, giving the same number and size of colonies as wild type, or gave no colonies at the highest arabinose.
All eight of the bottom mutants in the synergy loop failed to complement, indicating a loss of cell division function. The triple mutant RLD271-273AKK failed to complement, although the single mutant D273K did complement. Five of the 13 top mutants complemented the null. The remaining eight top mutants all failed to complement, and these were selected for further study.
It was important to confirm that the mutant proteins were expressed in the complementation system. To confirm this, we diluted overnight cultures of each of the strains and grew them for 3 h at 42°C with 0.05% arabinose to induce expression. We then pelleted the bacteria and loaded equal amounts of whole bacterial lysates (by total protein concentration) on sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Western blotting confirmed that all mutants were expressed at a level of more than twice that of wild type (data not shown). The expression level is probably equal to the four- to fivefold overexpression seen in strain BW27783 (see below), but we used horseradish peroxidase staining here, which is not as quantitative as the Alexa Fluor used below.
Two of the mutants, E138K and R142D, were insoluble when overexpressed from pET. We investigated their solubility when expressed from pJSB2 by growing JKD7-1/pKD3 cells containing these mutations under maximal induction (0.05% arabinose) and fractionating cell lysates into soluble and insoluble fractions. Both E138K and R142D were found in the soluble and insoluble fractions at approximately equal levels (data not shown). When cells expressing wild-type FtsZ from pJSB2 were fractionated in this manner, approximately two-thirds of the FtsZ was soluble. We conclude that these mutants are expressed and soluble in the complementation test cells. Most of the other mutant proteins could be prepared as soluble proteins for in vitro studies, so we conclude these are also soluble under complementation conditions.
We also checked FtsZ with either an N- or C-terminal His tag. The N-terminal His tag protein assembled in vitro into protofilament pairs or small bundles, instead of single protofilaments. The C-terminal His tag sometimes formed large aggregates of spiral bundles, but after 1 min these resolved into normal protofilaments. In spite of these aberrations, both His tag proteins complemented the ftsZ null allele.
We checked for complementation in liquid culture by using a serial dilution log-phase growth assay, where the culture was diluted every 2 h to maintain log phase (36). Arabinose was added to 0.05%, and cells were grown at 42°C. Complementation was scored as the ability to maintain growth over several hours. The results agreed with the plate complementation assay with one exception. N186K, which failed to complement on plates, grew for at least 15 generations under serial dilution conditions. However, we found that when cultures were diluted more than 10,000 they failed to grow. This was not due to a titratable factor in the medium, as cultures that were diluted into medium filtered from growing log-phase cultures also failed to grow. We do not have an explanation for this phenomenon.
We note that the complementation test was performed at 42°C, and so temperature sensitive mutants would be scored as noncomplementing. Temperature-sensitive alleles cannot be detected when temperature is used to turn off the (essential) wild-type gene.
Dominant negative effects. We next tested the ability of these mutated FtsZ proteins to interfere with wild-type FtsZ function. We were particularly interested in determining what level of mutant expression relative to wild type was needed to produce a dominant negative effect. This required regulating the expression of the mutant protein over known levels relative to wild type. A problem with using the pBAD vector for regulated expression is that Escherichia coli has a feedback system for arabinose transport that results in an all-or-nothing response. As the arabinose concentration is increased, an increasing number of cells turn on arabinose transport and expression from the araBAD promoter, but each activated cell is maximally induced. Khlebnikov et al. (14) created cell lines in which the arabinose transporter araE is constitutively expressed. In the BW27783 strain, individual cells respond more homogeneously by increasing expression levels as arabinose is increased. For our dominant negative experiments we used BW27783 transformed with our pBAD-based complementation plasmid (pJSB2) carrying ftsZ with the mutation of interest.
To quantitate the expression level of FtsZ from pJSB2, we constructed a reporter plasmid that had an extended myc tag on the C terminus of wild-type FtsZ. The addition of 20 amino acids raised the molecular mass of the expressed protein approximately 2,100 Da, allowing it to be resolved from wild-type FtsZ in Western blots stained with anti-FtsZ antibodies. We found that the expression level of FtsZ-myc varied from about 1/10 to five times the wild-type level over the range of arabinose tested (Fig. 2). Importantly, at 0.00015% arabinose the mutant protein was expressed at about the same level as wild-type FtsZ. Expression then jumped to four to five times the wild type at higher arabinose levels. We also constructed a version of this plasmid (pJSB7) by replacing our modified Shine-Dalgarno sequence with a sequence more closely matching the consensus sequence. This gave somewhat higher expression levels at the higher arabinose concentrations, up to seven times the wild-type level. In a separate experiment (data not shown), Western blotting showed that all mutant proteins were expressed at the same level as wild-type FtsZ (from pJSB2) at 0.05% arabinose, and we assume that this applies to all induction levels. We also determined the absolute expression levels per cell for endogenous FtsZ in BW27783. This FtsZ varied from 10,000 to 14,000 molecules per cell, somewhat lower than the 15,000 we determined for strain BL21 (a B/r A strain) (17), but significantly higher than the 3,200 determined for E. coli B/r K (31).
Each mutant was tested for dominant negative effects over a range of arabinose induction. Table 1 gives the level of FtsZ expression at which an effect was first seen (i.e., the level at which colonies were visibly smaller than wild type) and the expression level that resulted in complete inhibition of colony formation after an 18-h incubation. None of the mutants produced dominant negative effects at a substoichiometric level. Five of the bottom mutants (N207C, D209K, F210A, D212C, and RLD271-3AKK) reduced colony size at expression levels equal to or approximately fourfold higher than wild-type FtsZ. Three of these mutants (D209K, F210A, and RLD271-3AKK) were totally dominant negative, completely preventing cell division and colony growth. This lethality required expression at 4 to 5 times the wild-type level. Overexpression of wild-type E. coli FtsZ had no effect on colony number or size at the maximal expression from either pJSB2 or pJSB7.
Several studies have examined the effects of expressing FtsZ from foreign bacteria in E. coli (19, 32, 37). Foreign FtsZ can colocalize at the Z-ring when expressed at low levels and is toxic when expressed at higher levels. However, the level of expression that produced toxicity was not determined in these previous studies. We examined this question using our dominant negative expression system, to see if foreign FtsZ might poison substoichiometrically. FtsZ proteins from Synechocystis and B. subtilis were similar to bottom mutants in dominant negative effects. FtsZ from M. pneumoniae had no effect, probably due to its high degree of sequence divergence. We assumed that these foreign FtsZs were expressed at the same level as E. coli FtsZ at the same arabinose level, but we were not able to determine their expression levels because the antibody gives only partial cross-reactivity.
The effective level of arabinose might be different in liquid culture and agar plates, so we tested the bottom mutants for dominant negative effects in liquid culture. Overnight cultures were diluted 1/10,000 into medium with different levels of arabinose, and observed several hours later as their OD600 reached levels of 0.2 to 0.6. For D209K and F210A, turbidity was visibly reduced at 0.0005% arabinose (the same level that caused smaller colonies) and turbidity was eliminated at 0.0015 and 0.05% arabinose (the levels that completely inhibited colony formation) (Table 1). We conclude that the expression level is the same in liquid culture and plates. For the less deleterious mutants, which only showed reduced colony size in the plate assay, the liquid culture assay was less sensitive and did not demonstrate a clear dominant effect.
We should note one caveat to the quantitation. Several of the mutants showed major changes in phenotype from 0.0005 to 0.005 or 0.05% arabinose (Table 1), where Fig. 2 suggests the FtsZ level changes only from three to five times the wild-type level. This suggests that the effective level of FtsZ may be undergoing larger changes in this range than indicated by the graph.
Remarkably, none of the top mutants (except D45A) showed any dominant effects in the colony assay. Two top mutants were tested in the pJSB7 vector, which has an improved Shine-Dalgarno sequence and an expression level up to seven times the wild type. G103K and R142D both gave smaller colonies at the highest expression levels, while expression of wild-type FtsZ at this level did not affect colony size. We conclude that while the top mutants are not completely innocuous, they are significantly weaker in effect than the bottom mutants.
Structure of mutant Z-rings. We investigated cell morphology and Z-ring structure under both complementation and dominant negative conditions. In both cases the cells were induced with 0.05% arabinose to give maximal induction of the mutant FtsZ. Mutant strains which complemented were typically longer than wild-type cells, but never more than four times as long. These mutants had some apparently normal Z-rings and also had FtsZ structures that appeared as partial rings or patches (Fig. 3, top row). We did not see spiral Z-rings in the complementing mutants. While the structures in these cells were not as sharp as the Z-rings in wild-type cells, they are apparently capable of performing cell division.
|
Several other noncomplementing mutants formed long cells that contained helical FtsZ structures. These included the previously described D212A and D212G (36), as well as Q47K, D209C, F210A, and D212C. Q47K formed the most dramatic helical structures, with a wide and variable pitch and often running throughout the cell (Fig. 3). D209C was unable to form any medial FtsZ structure, instead forming bright narrow-pitched spirals at the poles of the cell (Fig. 3). F210A formed rings, short two-turn spirals of narrow pitch, and bright FtsZ patches (Fig. 3). These structures appeared to be located at every other internucleoid space. D212C formed spirals of constant pitch and intensity that ran the length of the cell. These were similar to the spirals formed in D212A and D212G.
We also examined the structures formed under dominant negative conditions (Fig. 4). In most cases the structures were similar to those in the complementing cells, which is not surprising since the mutant protein is about five times the concentration of the wild-type FtsZ. Top mutants grown under these conditions had short cells, as would be expected since none of these mutants were dominant negative. (D45A has a dominant effect at high expression, but this effect seemed to be due to growth arrest of short cells.) Some of the top mutants formed essentially normal FtsZ structures, such as E138K, but there was a high background of diffuse patches (Fig. 4). Other top mutants formed spiral FtsZ structures. The Q47K spirals look very similar to the spirals formed in complementing conditions. We cannot tell if there are normal Z-rings present that are obscured by the spirals or if the spirals themselves can function and achieve division. The bottom mutants had a wider range of structural phenotypes under dominant negative conditions. D209C, the only bottom mutant that had no dominant effects, formed normal Z-rings in a patchy background. In contrast, D209A formed small bright patches that often failed to extend across the diameter of the cell and seemed to arrest cell growth. D209K, which had the strongest dominant negative effects, caused cell filamentation with bright patches and diffuse staining. D212G and F210A formed helical structures.
|
GTPase activity. We measured the binding of GTP to wild-type and mutant proteins, as summarized in Table 1. Wild-type FtsZ bound 1 mol of GTP per mol of protein, as did several of the mutants. Other mutants showed a significant reduction in GTP binding.
The GTPase of wild-type FtsZ is known to exhibit a Cc around 1 µM, below which there is very low GTPase and above which the GTPase is directly proportional to FtsZ concentration (13, 30, 37). We confirmed that wild-type FtsZ in HMK buffer had a Cc for hydrolysis of 1.0 µM and an activity (slope) of 14 GTP per FtsZ per min above the Cc (Fig. 5; Table 1). This activity was found for several preparations of wild-type FtsZ, but some preparations gave an activity 20 to 30% lower. This may reflect the quality of protein preparations and suggests that activities of mutant proteins may also be variable within this range.
|
Several mutants showed an apparent reduction in Cc. L68W and G105S (FtsZ84) showed a Cc that was too close to zero to be measured, and an activity of about 4 to 5 GTP per FtsZ per min. We have confirmed by assembly studies that L68W has a Cc of 0.1 µM in HMK buffer, 10-fold lower than that of wild-type FtsZ (3). Several of the mutants showed a low hydrolysis activity of 0.15 to 0.2 GTP per FtsZ per min. We believe this activity is due to contaminating GTPase enzymes. We expressed and purified an unrelated inert protein by the same purification protocol used for FtsZ, and the end product showed a GTPase activity similar to or of higher level than the low GTPase FtsZ mutant proteins. We believe that the mutants that show activity higher than this level (for example, an activity of 0.35 to 0.5 GTP per FtsZ per min) have a low level of hydrolysis of their own.
Since both top and bottom mutants presumably have a functional surface on the side opposite the mutation, we expected that a mixture of top and bottom mutants might form dimers with a normal protofilament interface and that these dimers might exhibit a GTPase activity. We tested the GTPase activity for selected pairs of top and bottom mutants (Fig. 6). The mixtures gave a GTPase that was no more than the sum of the two components, and in some cases somewhat less than additive. We thus found no evidence for GTPase activity that could be attributed to a top and bottom mutant dimer.
|
|
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This assumption appears to be contradicted by our observation that several of the bottom mutants had substantially reduced binding of GTP. How could a point mutation of a surface residue on the bottom be transmitted to the GTP binding site on the top? We think it is unlikely that the bottom mutants generate a structural change that is transmitted all the way to the GTP binding site. We note that both D209K and F210A, which showed the strongest block of GTP binding, exhibited strong dominant negative effects, implying that their top surface is functional and likely bound to GTP in the cell. One possibility is that GTP binding requires a dimeric structure, perhaps related to the dimers of FtsZ-GDP found under conditions of molecular crowding (12). The full explanation awaits further studies.
We attempted negative-stain electron microscopy (EM) to characterize the assembly of mutant proteins and their coassembly with wild-type FtsZ, but the results were not quantitative and often ambiguous. We then used our recently developed tryptophan fluorescence assay (3) to characterize selected mutants (Y. Chen and H. P. Erickson, unpublished observations). A double mutant, L68W/D209K, showed no fluorescence change upon GTP addition, either by itself at >10 µM or with wild-type FtsZ. Thus, it cannot assemble. D209A and D209C coassembled with L68W, consistent with previous studies (18, 33). The top mutant Q47K also coassembled with L68W, consistent with the prominent spiral polymers seen in bacteria by immunofluorescence (Fig. 3 and 4). This limited survey confirms that some of our mutants are coassembly mutants and some are cap mutants, as discussed next.
There are two ways that a mutant protein could poison wild-type FtsZ and produce dominant negative effects. The first is by coassembly to produce an aberrant polymer, and the second is by capping polymers and/or sequestering monomers. A precedent for the first is the study of Anders et al. (1), who tested the synergy loop mutant E254A of -tubulin (E255A in yeast), which corresponds to our FtsZ mutants D212A/C/G. The mutant tubulin coassembled with wild-type tubulin, and these microtubules were much more resistant to disassembly than normal. The tubulin E254A mutant substoichiometrically poisoned yeast microtubules when expressed at a level of only 1/20 that of wild-type -tubulin.
The second way to produce a dominant negative effect is by capping. An example of a microtubule capping protein is N-ethyl maleimide-modified tubulin. This modified tubulin was found to block growth of the minus end, while not affecting the plus end. It thus seems to act as a minus end cap. It also was effective at substoichiometric levels, in a range from 0.1 to 1.0 relative to wild-type tubulin (27). While this covalent modification results in a substoichiometric cap, mutation of the proteins might not necessarily have this effect. In comprehensive studies of mutants in yeast - and ß-tubulin, only 2 out of 49 -tubulin (29) and 2 out of 46 ß-tubulin (28) mutant clusters were dominant negative. The tubulin mutants were only tested at a 1:1 stoichiometry, a ratio where we saw no dominant effects for any of our FtsZ mutants. -Tubulin E254A may be a rare mutant with substoichiometric dominant negative effects.
Bottom mutants.
Several of our bottom mutants (D209A/C and D212A/C/G) are likely to be coassembly mutants, since they have been reported to assemble on their own or to coassemble with wild-type FtsZ in vitro (18, 34). However, in contrast to the substoichiometric poisoning seen in -tubulin E254A, these mutants only inhibited cell growth when expressed at four to five times the level of wild-type FtsZ. This suggests that FtsZ protofilaments can function for cell division with a majority of the interfaces being mutated and incapable of GTP hydrolysis. These putative coassembly mutants were the least deleterious of the bottom mutants. They reduced colony size but not colony number at the highest expression level, 5 times wild type. Several bottom mutants are likely cap mutants: N207C, D209K, F210A, and RLD271-3AKK. These mutants appear to be more toxic, since they completely block colony formation at the higher expression levels.
It is probably more meaningful to compare the level of the mutants to that of the wild-type monomer, rather than to total FtsZ. Total FtsZ was measured to be 11 µM in strain BL21 (17) and 7 to 10 µM in strain BW27783 (present study). If the Cc in the cell is 1 µM, which was determined in vitro under physiological buffer conditions (3, 12), the wild-type FtsZ should have 6 to 9 µM subunits in protofilaments and 1 µM free subunits. (This assumes that the 70% of the FtsZ that is in the cytoplasm is also mostly assembled into protofilaments, which we think is likely but have not yet demonstrated.) A bottom cap mutant present at 4 to 5 times the total FtsZ would actually have an effective concentration 25 to 50 times the concentration of free wild-type monomers. It seems that these bottom cap mutants are more than an order of magnitude weaker than wild-type FtsZ in competing for protofilament assembly.
The cell division inhibitor SulA binds to the bottom interface of FtsZ and probably blocks assembly of FtsZ by acting as a steric block of that interface (5). It has been suggested that SulA may act substoichiometrically by capping FtsZ protofilaments (5). There is also one report that SulA does indeed poison E. coli substoichiometrically (20). However, we would suggest that the FtsZ-SulA complex should act like a bottom cap mutant. If so it would interact only weakly with protofilaments and would have to act stoichiometrically by binding and sequestering FtsZ subunits. It should be interesting to reinvestigate the stoichiometry of SulA to FtsZ in vivo and in vitro.
Top cap mutants are even weaker, possible treadmilling. Even more surprising than the weak activity of the bottom mutants are the top mutants. They showed little or no dominant negative effect even at the highest expression level. This is 5 times the level of wild-type FtsZ, or 50 times the 1 µM Cc. A sevenfold overexpression using pJSB7 caused reduced colony size for some mutants but did not completely block growth. Consistent with the lack of activity in vivo, a fivefold excess of top mutant had no effect on wild-type GTPase activity. Several of the top mutants produced aberrant Z-ring morphology, even though they did not reduce colony size. In a related study, the top mutant T108A (FtsZ3) caused Z-ring aberrations when overexpressed (2). Therefore, the top mutants are not completely innocuous but are significantly weaker in effect than the bottom mutants.
The weak activity of the top mutants suggests the possibility that assembly of protofilaments may be directional. If subunits add primarily at the bottom end, and dissociate primarily from the top, top mutants may be inactive because there is no significant assembly at the top end. Assembly at one end and disassembly at the other requires an irreversible reaction step like GTP hydrolysis (38) and is termed treadmilling.
Interface mutants and cooperativity.
Although the N- and C-terminal globular domains of FtsZ appear to be independently folding domains (25), the present study suggests that their assembly surfaces do not act independently. If one interface surface is missing or damaged, the other has a much weaker interaction than predicted by the 1 µM critical concentration (which is equal to the Kd for elongation). Our results suggest that the subunit needs a functional interface at both the top and bottom in order to interact with protofilaments at the 1 µM Kd. At the present time we cannot propose a structural basis for this speculation, but it may be related to the enigma of the apparent cooperativity of FtsZ assembly (3). We may require a model of the protofilament structure that is very different from the simple stacking of subunits on top of each other.
|
ACKNOWLEDGMENTS |
|---|
The work was supported by NIH grant GM66014.
|
FOOTNOTES |
|---|
Present address: Program in Molecular Medicine, University of Massachusetts Medical School, Worcester, MA 01605.
Present address: Department of Bioengineering, University of California, San Diego, La Jolla, CA 92093-0412.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
ß tubulin dimer by electron crystallography. Nature 391:199-203.[CrossRef][Medline]
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Appl. Environ. Microbiol. | Infect. Immun. | Eukaryot. Cell |
|---|---|---|
| Mol. Cell. Biol. | J. Virol. | Microbiol. Mol. Biol. Rev. |
| ALL ASM JOURNALS |
