Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. 2002 Mar 5; 99(5): 3171–3175.
Published online 2002 Feb 19. doi:  10.1073/pnas.052595099
PMCID: PMC122491

Rapid assembly dynamics of the Escherichia coli FtsZ-ring demonstrated by fluorescence recovery after photobleaching


FtsZ, the major cytoskeletal component of the bacterial cell-division machine, assembles into a ring (the Z-ring) that contracts at septation. FtsZ is a bacterial homolog of tubulin, with similar tertiary structure, GTP hydrolysis, and in vitro assembly. We used green fluorescent protein-labeled FtsZ and fluorescence recovery after photobleaching to show that the E. coli Z-ring is extremely dynamic, continually remodeling itself with a half-time of 30 s. ZipA, a membrane protein involved in cell division that colocalizes with FtsZ, was equally dynamic. The Z-ring of the mutant ftsZ84, which has 1/10 the guanosine triphosphatase activity of wild-type FtsZ in vitro, showed a 9-fold slower turnover in vivo. This finding implies that assembly dynamics are determined primarily by GTP hydrolysis. Despite the greatly reduced assembly dynamics, the ftsZ84 cells divide with a normal cell-cycle time.

The processes involved in bacterial cell division are strikingly different from those involved in eukaryotic cytokinesis. Whereas animal cells use a contractile ring of actin and myosin, bacteria form a ring of FtsZ, a cell division protein found in almost every prokaryotic species. This ring, known as the Z-ring, is attached to the inner membrane and contracts during septation. FtsZ is the major protein in the ring, although other proteins necessary for division have been identified in Escherichia coli and other species (13). FtsZ is a homolog of tubulin: both proteins share very similar tertiary structures (4), hydrolyze GTP, and assemble into straight and curved protofilaments (58).

Although the substructure of the Z-ring is not known, in vitro studies suggest that it is based on protofilaments, perhaps associated into multistranded structures. FtsZ polymerizes into single protofilaments under many in vitro solution conditions. Under other conditions, the protofilaments associate side-by-side to form ribbons or sheets (5, 7, 9). An assembly mixture of single protofilaments hydrolyzes GTP at a rate of five molecules per min per FtsZ (10) in physiological buffer conditions (350 mM KCl, pH 7.5; ref. 11). There is no evidence that FtsZ undergoes dynamic microtubule-like instability. Instead, a model of isodesmic assembly has been proposed in which each GTP hydrolysis event results in fragmentation of the protofilament, followed by nucleotide exchange and reannealing (12). In this model, the guanosine triphosphatase (GTPase) activity would reflect very rapid fragmentation/reannealing dynamics, in which every subunit is turned over, in the sense of breaking and reforming an interface, five times per minute at steady-state.

In rapidly dividing E. coli cells, the Z-ring forms in the center of the cell about 1 to 5 min after division, remains for 15 min, and then quickly constricts to divide the cell (1315). Although the Z-ring appears static when observed by fluorescence of FtsZ-gfp, it could be exchanging subunits with a cytoplasmic pool. In an attempt to directly investigate the assembly dynamics of FtsZ, we performed fluorescence recovery after photobleaching (FRAP) on Z-rings containing a small proportion of FtsZ-gfp.

Materials and Methods

Strains and Plasmids.

MC1000 (16) was used as a wild-type strain. JFL101 [ftsZ84(ts), supplied by J. Lutkenhaus, University of Kansas Medical Center] (17) grows normally at 30°, but its Z-rings disassemble, and it grows into filaments at 42°. PB103(λCH50) (zipA-gfp, supplied by P. de Boer, Case Western Reserve University, Cleveland, OH) (18) carries a stable phage containing zipA-gfp under control of the lac promoter. LMG64 [ftsI(ts), supplied by D. S. Weiss, Harvard Medical School] (19), grows normally at 30°, but grows into filaments with multiple Z-rings at 42°.

pBZG is based on pBAD18 plasmid (20) and expresses a fusion protein consisting of E. coli FtsZ, an HIIH linker, mut2 green fluorescent protein (gfp), and a 6xHis C-terminal tag under the control of the pBAD promoter. The junctions of this plasmid were confirmed by sequencing. pBZ84G was derived from pBZG by mutating the ftsZ to ftsZ84 (G105S) by using the QuickChange kit (Stratagene), following manufacturer's directions and confirming the mutation by sequencing.

Growth and Induction.

All strains were grown in Mops+ medium (supplemented with amino acids, vitamins, and nucleic acids) as in ref. 21 to avoid the background fluorescence associated with casamino acids. Ampicillin was added to growth media for transformed strains at a concentration of 100 μg/ml. Either glucose or glycerol (0.2%) was added as a carbon source to repress or allow, respectively, expression of FtsZ-gfp. MC1000(pBZG) or LMG64(pBZG) were grown in Mops+ containing 0.2% glucose at 37°. For microscopy and FRAP, overnight cultures were diluted into Mops+ medium containing 0.2% glycerol and grown at 37° to A600 = 0.5, then induced by the addition of arabinose to 0.00005% and grown at 37° for 45 min before being transferred back into fresh Mops+ medium with glycerol and no arabinose (preventing further FtsZ-gfp expression) for viewing.

JFL101(pBZ84G) was grown in Mops+ containing 0.2% glucose at 30°. For microscopy and FRAP, overnight cultures were diluted into Mops+ containing 0.2% glycerol and 0.00001% arabinose and grown at 30° to A600 = 0.5 (approximately 15 h), then transferred to fresh Mops+ containing 0.2% glycerol and no arabinose for viewing. Induction in JFL101(pBZ84G) was less reproducible than induction in MC1000(pBZG), necessitating longer growth with lower concentrations of arabinose. The doubling time of JFL101(pBZ84G) did not change upon induction, and cells with normal Z-rings were common. PB103(λCH50) was grown and induced as in ref. 18.

Microscopy and FRAP.

Image acquisition and FRAP measurements were performed on a Nikon TE300 fluorescence microscope system with a temperature-controlled stage as described (22). Briefly, cells were suspended in Mops+ medium with glycerol, attached to polylysine-coated coverslips, and then imaged through a 100×/1.4 numerical aperture Plan Apochromat objective lens with an Orca cooled charge-coupled device camera (Hamamatsu Photonics, Hamamatsu, Japan). All observations were performed at 30°C. Exposures were controlled by METAMORPH software (Universal Imaging, Media, PA). Photobleaching was achieved by focusing an argon ion laser to a diffraction-limited spot on the specimen for an exposure of ≈25 ms. Wide-field fluorescent images of the cells were acquired before and after photobleaching by custom time-lapse recording of digital images with 12-bit gray levels. Images of MC1000(pBZG) cells and PB103(λCH50) cells were obtained every 10–15 s for 2 min, after which the time-lapse interval was increased to 2 min. Images of JFL101(pBZ84G) cells were obtained with a time-lapse interval of 2–3 min. Measurements of integrated fluorescence in regions of the cell as well as calculation of the half-time and percent fluorescence recovery in the 0.5 μm wide photobleached region of rings were performed as described (22).


E. coli MC1000 cells were transformed with pBZG, an inducible plasmid expressing FtsZ-gfp, and induced such that Z-rings were visible by fluorescence microscopy (Fig. (Fig.11 Upper Left). The morphology of these rings was similar to that of Z-rings visualized by immunofluorescence microscopy, and cell growth and division were unaffected. Western blot analysis showed that less than 25% of the total FtsZ in the cell was FtsZ-gfp (data not shown).

Figure 1
FRAP analysis of FtsZ. MC1000(pBZG) cells were visualized on the microscope slide. Half of the Z-ring of the cell on the left was laser photobleached at the arrow, and recovery of fluorescence was monitored. Panels were photographed at times indicated ...

Analysis of the distribution of FtsZ-gfp in the cells by integrating fluorescence over the ring and the cytoplasm showed that only ≈30% of the FtsZ in the cell is incorporated into the Z-ring. The remainder of the FtsZ is diffused throughout the cell, presumably as free cytoplasmic subunits or small assemblies. This result means that at any given time, there is a large soluble pool of FtsZ available to exchange into the ring. It also means that the Z-ring is, on average, only six to seven protofilaments thick (see ref. 10 for this calculation), perhaps explaining why the Z-ring has never been seen by thin-section electron microscopy.

To investigate whether exchange occurs between the Z-ring and the cytoplasmic pool, we performed FRAP analysis on the gfp-labeled Z-rings. FRAP uses a focused laser beam to photobleach a portion of a fluorescent structure and measures the recovery of fluorescent signal in the photobleached area caused by exchange of subunits from outside of the photobleached area. Z-rings recovered fluorescence very quickly (Fig. (Fig.1),1), with a mean halftime of recovery of 31.7 ± 3.4 s (mean ± SE, n = 13) at 30°. Z-rings that were entirely photobleached did not fully recover fluorescence because of the lower pool of fluorescent subunits available after bleaching. However, Z-rings that had been only partially bleached usually recovered fluorescence to nearly prephotobleaching levels.

We also performed FRAP on LMG64(pBZG) cells (23), which are ftsI(ts) and cannot divide at the restrictive temperature but can still elongate. These cells contain multiple fluorescent Z-rings and a larger total pool of available FtsZ-gfp subunits. When a single Z-ring was repeatedly bleached, it recovered fully with the same ≈30 s half life as in the wild-type cells (data not shown).

After photobleaching, fluorescence always reappeared in structures identical to those seen before photobleaching, even in the case of rare aberrant FtsZ structures such as helices or punctate rings. To rule out the possibility that recovery of fluorescence was caused by recovery of gfp fluorophores, we fixed cells with 4% (wt/vol) paraformaldehyde and measured their FRAP recovery rate. The Z-rings appeared normal but showed no recovery of fluorescence into the photobleached area, whether at the Z-ring or in the cytoplasm, after 20 min (data not shown). This result demonstrates that recovery of fluorescence depended on exchange of fluorescent FtsZ subunits from the cytoplasmic pool. The photobleaching procedure was not harmful to the cells or the rings; completely photobleached cells grew and divided on the slide in a manner similar to that of adjacent unbleached cells (data not shown).

We considered the possibility that the dynamic behavior of the Z-ring is required for its initial assembly and localization, but that during constriction, the Z-ring becomes a stable scaffold for a force-generating mechanism. However, we found no relationship between cell length (used as a marker for progress into the cell cycle in the logarithmic phase cells being analyzed) and halftime of recovery (data not shown). FRAP analysis of cells with visible division furrows (Fig. (Fig.2)2) showed a halftime of recovery of 35.6 ± 14.2 s (mean ± SE, n = 10). This rate of recovery is similar to that of nondividing cells, indicating that the Z-ring is dynamic during constriction. Intriguingly, some dividing cells had ≈60% of their FtsZ incorporated into the Z-ring, twice the proportion in the Z-ring of nondividing cells; however, this increase in incorporation into the Z-ring was not consistently observed.

Figure 2
FRAP analysis of the Z-ring in a dividing cell. (Upper Left) Brightfield image of an MC1000(pBZG) cell in the process of division. The cell was photobleached at the arrow and the recovery of fluorescence was monitored. Panels were photographed at times ...

The dynamics and exchange of subunits in the microtubule cytoskeleton are coupled to GTP hydrolysis. FtsZ hydrolyzes GTP at a rate of 5 min−1 in vitro (10), a rate similar to the exchange in vivo. We tested the possible coupling of FtsZ exchange to GTP hydrolysis with an E. coli strain (JFL101) that has the mutant ftsZ84 allele at the genomic locus (17). This mutant FtsZ has ≈10% of the GTPase activity of wild-type FtsZ when measured in vitro (24, 25). Despite this greatly reduced GTPase, cells carrying ftsZ84 seem to divide normally at 30°, with the same cell-cycle timing as wild-type cells. We inserted the ftsZ84 mutation (G105S) into our ftsZ-gfp plasmid to create pBZ84G and transformed JFL101 cells with this mutated plasmid. Thus, both the genomic and the gfp-labeled FtsZ carried the ftsZ84 mutation. Although FtsZ84-gfp induction in JFL101(pBZ84G) was less reproducible, we found conditions in which many of the cells contained a single bright medial ring of normal appearance, and we performed FRAP analysis on these rings.

Quantitative fluorescence analysis showed that in these ftsZ84 cells, the Z-rings contained ≈65% of the FtsZ in the cell, a proportion twice as high as that found for wild-type FtsZ. In our FRAP studies, Z-rings consisting of FtsZ84 recovered much more slowly than wild-type Z-rings (Fig. (Fig.33 and Fig. Fig.4).4). The halftime of recovery was 284.0 ± 19.5 s (mean ± SE, n = 12) at 30°C, ≈9-fold longer than that of wild-type rings. This finding corresponds very closely to the difference in in vitro GTPase rate and suggests that the turnover of FtsZ in vivo is tightly coupled to GTP hydrolysis. This fact may also explain the high proportion of FtsZ84 associated with the ring. If turnover and release from the ring are coupled to GTP hydrolysis, but incorporation into the ring is not, then FtsZ that has a slower rate of hydrolysis might accumulate in the ring.

Figure 3
FRAP analysis of FtsZ84. JFL101(pBZ84G) cells were visualized on the microscope slide. Two cells are outlined in Upper Left. The entire Z-ring of the cell at Upper Right was laser photobleached as indicated by the arrow, and recovery of fluorescence was ...
Figure 4
Comparison of recovery of fluorescence of wild-type FtsZ and FtsZ84. Data point values are normalized to the expected final recovery of each species of FtsZ, determined by assuming an exponential recovery with the halftimes given in the text and taking ...

In E. coli, the transmembrane protein ZipA, an essential component of the division machinery, binds FtsZ and may serve as a membrane tether (18). If the ZipA ring were stable, then it might act as a scaffold for FtsZ accumulation. Even if it were connected to the Z-ring only through its attachment to FtsZ, its turnover and exchange might be slowed because it is confined to the membrane. We used a zipA-gfp strain, PB103(λCH50), to determine the assembly dynamics of ZipA in the Z-ring. Approximately 30% of the ZipA in the cell was incorporated into a medial ring, a proportion similar to that seen with FtsZ. The ZipA ring appeared to be as dynamic as FtsZ (Fig. (Fig.5),5), with a recovery halftime of 27.9 ± 6.0 s (mean ± SE, n = 9). It is possible that the entire division machinery undergoes a similar dynamic exchange.

Figure 5
FRAP analysis of ZipA. (A) PB103(λCH50) cells were laser photobleached at the ZipA ring (arrow); recovery of fluorescence was monitored over time. (B) A PB103(λCH50) cell was attached to the coverslip at its end, allowing a face view of ...


The quick recovery of fluorescence that we observe implies that the Z-ring is an extremely dynamic structure, continually exchanging subunits with a cytoplasmic pool. The halftime of recovery seen with FtsZ is roughly comparable to the most dynamic cytoskeletal structures. Spindle microtubule arrays show a halftime of recovery of ≈10–20 s (26, 27), whereas other, more stable structures recover much more slowly; such structures include interphase microtubule arrays, halftime of recovery of 200 s (27), actin stress fibers, halftime of recovery of 500 s (28), or vimentin, halftime of recovery of 10–15 min (29). Yeast metaphase mitotic spindles show a halftime of recovery of ≈1 min, similar to that seen here (22).

Two previous studies have already demonstrated rapid dynamics of assembly during the periods of active assembly or disassembly of the Z-ring. Sun and Margolin (14) found that FtsZ rings assembled in daughter cells very soon after the parental Z-ring constricted and disassembled. Significantly, it took only ≈1–3 min for the new rings to go from very dim to full brightness. Our work now extends this picture of rapid initial assembly to show that once the ring has reached its steady-state, it continues to exchange subunits and remodel at least as fast as the initial assembly. Presumably, this exchange also occurs as the ring is initially assembled. In the second study, Addinall et al. (30) found that Z-rings of FtsZ84 disassembled completely in 1 min when shifted to 42°C, which is much faster than the ≈5 min half time for remodeling we observed at 30°C. Below, we suggest that remodeling within the ring may be much faster than exchange with the cytoplasmic pool, so this rapid disassembly may indicate the total rate of fragmentation when reannealing is blocked. When shifted from 42° back to 30°C, the Z-rings reassembled over a period of 5 min, which is similar to the remodeling rate we observed.

In any study using gfp-labeled proteins, it is possible that labeled protein may behave differently from unlabeled. Previous studies have concluded that FtsZ-gfp coassembles with the endogenous FtsZ to label the Z-ring, is at least partially functional, and causes no aberrations when expressed at less than 30% of the normal level (13, 14). It is possible that a decreased activity of FtsZ-gfp is the reason that only 30% of FtsZ-gfp was in the Z-ring. However, the fact that 65% of FtsZ84-gfp localizes to the Z-ring suggests that this is not the case. Although it is difficult to prove conclusively, we think it is likely that the localization and dynamics of FtsZ-gfp reflect that of the native FtsZ.

If FtsZ in vivo is hydrolyzing GTP at the rate of five molecules per min per FtsZ (the rate measured in vitro for physiological conditions; ref. 10), the 15,000 molecules in an average E. coli cell would be consuming 1,250 molecules of GTP per s. This hydrolysis may seem to be a significant energetic burden to recycle subunits into the Z-ring, but it is actually an insignificant fraction of the cell's metabolism. Actively growing E. coli have been measured to consume 6.2 × 106 molecules of O2 per s, which would produce 3.7 × 107 high-energy phosphates per s (calculated assuming three phosphates per oxygen atom; ref. 31). Thus, the five GTPs hydrolyzed per min per FtsZ represent only 1/30,000 of the active cell's metabolism.

Although little is known about the structure and assembly of the Z-ring in vivo, much has been learned about FtsZ polymers from in vitro assembly studies. Depending on assembly conditions, FtsZ can assemble into single protofilaments or small bundles and sheets of protofilaments. If FtsZ assembly is cooperative, like microtubules, one would postulate an association of FtsZ subunits onto the ends of multifilament bundles. In contrast to this, Romberg et al. (12) have described a theory of isodesmic assembly in which FtsZ subunits assemble into protofilaments one subunit wide. Assembly can occur by addition of single subunits or by annealing of filaments, and fragmentation can occur at any binding interface. In this model, all subunits in a protofilament are postulated to have bound GTP, and hydrolysis to GDP results in rapid fragmentation at that interface. The filament is able to reanneal after exchange of GDP for GTP. If the Z-ring is constructed from these isodesmic protofilaments, then we would expect to see a correlation between the GTPase activity of FtsZ and Z-ring stability.

Fig. Fig.66 presents two models for FtsZ distribution in the bacterium and Z-ring structure. The top model is based on cooperative assembly. Protofilament bundles are localized to the Z-ring, and FtsZ in the cytoplasm is monomeric. The rapid dynamics of the ring would be explained best by a mechanism similar to dynamic instability, in which whole segments of the polymers disassemble into monomers and are replaced by new segments assembled from monomers. The bottom model is based on the isodesmic assembly hypothesis. At the 11 μM concentration of FtsZ obtained in vivo (10), the model predicts that virtually all of the FtsZ in the cell will be assembled into protofilaments with an average length of ≈80 subunits (protofilaments averaging 25 subunits have been observed in vitro at a 10-fold lower FtsZ concentration; ref. 12). This model suggests that the pool of cytoplasmic FtsZ may consist of protofilaments rather than single subunits. A quantity of 15,000 molecules of FtsZ would give ≈180 protofilaments of 80 subunits. (More precisely, there would be a distribution around this length.) Thirty percent, or ≈56 protofilaments, would be concentrated in the Z-ring, with the remainder distributed throughout the cytoplasm. In this model, the assembly dynamics of the Z-ring would involve the exchange of short pieces of protofilaments, released after GTP hydrolysis and fragmentation, rather than of single subunits. In either model, there are likely to be events during which FtsZ is released from the Z-ring but reassociates within the ring, rather than diffusing to the cytoplasm. These internal remodeling events are likely to exceed those resulting in net exchange, which implies that the total assembly dynamics of the Z-ring are faster than the exchange measured by FRAP.

Figure 6
Two models of how FtsZ protofilaments might be distributed in the Z-ring and cytoplasm. The semicircles are transverse cross sections of 200-nm thickness through the bacterium at the cytoplasm (Left) or Z-ring (Right), and the total number of subunits ...

The isodesmic assembly model also postulates a pathway for protofilament fragmentation that is independent of GTP hydrolysis. In the model developed from in vitro studies (12), this event would occur at a rate of 0.9 per min. However, the comparative dynamics of wild-type FtsZ and FtsZ84 suggest that the dynamics of the Z-ring are dominated by the GTPase-dependent mechanism, and that the hydrolysis-independent pathway does not play a major role in subunit turnover.

The remarkably dynamic nature of the Z-ring at first seems enigmatic either for its role as a scaffold or for generating a constriction force. In eukaryotes, the actin filaments provide a stable scaffold upon which myosin motors operate to generate the force of constriction. Whatever the mechanism for force generation in the Z-ring, it must be able to operate with the constant turnover of its scaffold. But we also should consider the possibility that the rapid turnover may be more than an obstacle for force generation. Fragmentation and annealing might be part of the mechanism for generating the force for constriction. For example, if a mechanism existed for sequestering FtsZ in the cytoplasm at the time of constriction, protofilament pieces might be removed from the ring more quickly than they are replaced. Reannealing of the remaining filaments could only occur when they move closer together, and this might lead to a circumferential constriction that could power cell division.


We would like to thank Tom Melby for construction of pBZG and Dr. Petra Levin and Dr. Laura Romberg for helpful comments on the manuscript. This study was funded by National Institutes of Health Grants GM28553 (to H.P.E.) and GM60678 (to E.D.S.).


fluorescence recovery after photobleaching
guanosine triphosphatase
green fluorescent protein


This paper was submitted directly (Track II) to the PNAS office.


1. Bramhill D. Annu Rev Cell Dev Biol. 1997;13:395–424. [PubMed]
2. Lutkenhaus J, Addinall S G. Annu Rev Biochem. 1997;66:93–116. [PubMed]
3. Rothfield L, Justice S, Garcia-Lara J. Annu Rev Genet. 1999;33:423–448. [PubMed]
4. Löwe J. J Struct Biol. 1998;124:235–243. [PubMed]
5. Erickson H P, Taylor D W, Taylor K A, Bramhill D. Proc Natl Acad Sci USA. 1996;93:519–523. [PMC free article] [PubMed]
6. Lu C L, Reedy M, Erickson H P. J Bacteriol. 2000;182:164–170. [PMC free article] [PubMed]
7. Löwe J, Amos L A. EMBO J. 1999;18:2364–2371. [PMC free article] [PubMed]
8. Löwe J, Amos L A. Biol Chem. 2000;381:993–999. [PubMed]
9. Yu X C, Margolin W. EMBO J. 1997;16:5455–5463. [PMC free article] [PubMed]
10. Lu C, Stricker J, Erickson H P. Cell Motil Cytoskeleton. 1998;40:71–86. [PubMed]
11. Cayley S, Lewis B A, Guttman H J, Record M T., Jr J Mol Biol. 1991;222:281–300. [PubMed]
12. Romberg L, Simon M, Erickson H P. J Biol Chem. 2001;276:11743–11753. [PubMed]
13. Ma X, Ehrhardt D W, Margolin W. Proc Natl Acad Sci USA. 1996;93:12998–13003. [PMC free article] [PubMed]
14. Sun Q, Margolin W. J Bacteriol. 1998;180:2050–2056. [PMC free article] [PubMed]
15. Den Blaauwen T, Buddelmeijer N, Aarsman M E, Hameete C M, Nanninga N. J Bacteriol. 1999;181:5167–5175. [PMC free article] [PubMed]
16. Casadaban M J, Cohen S N. J Mol Biol. 1980;138:179–207. [PubMed]
17. Lutkenhaus J, Wolf-Watz H, Donachie W D. J Bacteriol. 1980;142:615–620. [PMC free article] [PubMed]
18. Hale C A, De Boer P A J. Cell. 1997;88:175–185. [PubMed]
19. Guzman L M, Weiss D S, Beckwith J. J Bacteriol. 1997;179:5094–5103. [PMC free article] [PubMed]
20. Guzman L M, Belin D, Carson M J, Beckwith J. J Bacteriol. 1995;177:4121–4130. [PMC free article] [PubMed]
21. Neidhardt F C, Bloch P L, Smith D F. J Bacteriol. 1974;119:736–747. [PMC free article] [PubMed]
22. Maddox P S, Bloom K S, Salmon E D. Nat Cell Biol. 2000;2:36–41. [PMC free article] [PubMed]
23. Weiss D S, Pogliano K, Carson M, Guzman L M, Fraipont C, Nguyen-Distèche M, Losick R, Beckwith J. Mol Microbiol. 1997;25:671–681. [PubMed]
24. RayChaudhuri D, Park J T. Nature (London) 1992;359:251–254. [PubMed]
25. de Boer P, Crossley R, Rothfield L. Nature (London) 1992;359:254–256. [PubMed]
26. Salmon E D, Leslie R J, Saxton W M, Karow M L, McIntosh J R. J Cell Biol. 1984;99:2165–2174. [PMC free article] [PubMed]
27. Saxton W M, Stemple D L, Leslie R J, Salmon E D, Zavortink M, McIntosh J R. J Cell Biol. 1984;99:2175–2186. [PMC free article] [PubMed]
28. Wang Y L. J Cell Biol. 1987;105:2811–2816. [PMC free article] [PubMed]
29. Vikstrom K L, Lim S-S, Goldman R D, Borisy G G. J Cell Biol. 1992;118:121–129. [PMC free article] [PubMed]
30. Addinall S G, Cao C, Lutkenhaus J. J Bacteriol. 1997;179:4277–4284. [PMC free article] [PubMed]
31. Imlay J A, Fridovich I. J Biol Chem. 1991;266:6957–6965. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...