Biophysical properties of stable microtubules in neurites revealed by optical techniques

We tested the stability of microtubules in the neurites of cultured dorsal root ganglion cells by dissolving the cytoplasmic membrane with detergent and exposing them to defined extracellular medium under the microscope. Smooth cytoplasmic filaments visualized after membrane removal were suggested to be microtubules by the preservation of all of the filaments in the presence but not in the absence of taxol. They were further confirmed to be microtubules by immunostaining with anti-tubulin antibody. Significant number of microtubules in the established neurites remained longer than 1 hour after membrane removal. To investigate their stabilization mechanism, we transected the exposed microtubules by laser microbeam irradiation and observed their length changes with video-enhanced microscopy. Microtubule fragments started to shorten on both sides of the transection site, more rapidly from the newly generated plus ends than from the minus ends. The maximal rate as well as the pattern of shortening correlated with the time of transection; microtubules transected later than 30 min after membrane removal shortened at rates less than 20 microm/min and typically with intermittent pauses, while the more labile microtubules included in the earlier transections shortened continuously at higher rates. Microtubules in neurites were thus stabilized by (1) stopping disassembly at local sites including the plus ends, and (2) slowing disassembly along the length. Observations of the course of disassembly also suggested the presence of specialized points along microtubules which is involved in anchoring microtubules to the substratum or transiently stopping disassembly.     

Effects of magnesium on the dynamic instability of individual microtubules

We investigated the effect of magnesium ion (Mg) on the parameters of dynamic instability of individual porcine brain microtubules. Rates of elongation and rapid shortening were measured by using video-enhanced DIC light microscopy and evaluated by using computer-generated plots of microtubule length vs time. Increasing [Mg] from 0.25 to 6 mM increased the second-order association rate constant for elongation about 25% at each end. At plus ends, this resulted in a 1.5-2-fold increase in elongation rates over the tubulin concentrations explored. Rapid shortening rates were more dramatically affected by Mg. As [Mg] was increased from 0.25 to 6 mM, the average rate of rapid shortening increased about 3-fold at plus ends and 4-5-fold at minus ends. The ends had roughly equivalent average rates at low [Mg], of 30-45 microns/min. At any Mg concentration, rates of disassembly varied from one microtubule to another, and often an individual microtubule would exhibit more than one rate during a single shortening phase. Individual rates at 6 mM Mg varied from 12 to 250 microns/min. Over the concentration range explored, Mg affected the frequencies of transition from elongation to shortening and back only at minus ends. Minus ends were relatively stable at low [Mg], having 4 times the frequency of rescue than at high [Mg], and a lower frequency of catastrophe (particularly evident at low tubulin concentrations). Plus ends, surprisingly, were highly unstable at all Mg concentrations investigated, having about the same transition frequencies as did the least stable (high Mg) minus ends. Our results have implications for models of the GTP cap, again emphasizing that GTP caps cannot build up in proportion to elongation rate, and must be constrained to the tips of growing microtubules.     

Dilution-induced disassembly of microtubules: relation to dynamic instability and the GTP cap

Microtubules were assembled from purified tubulin in the buffer originally used to study dynamic instability (100 mM PIPES, 2 mM EGTA, 1 mM magnesium, 0.2 mM GTP) and then diluted in the same buffer to study the rate of disassembly. Following a 15-fold dilution, microtubule polymer decreased linearly to about 20% of the starting value in 15 sec. We determined the length distribution of microtubules before dilution, and prepared computer simulations of polymer loss for different assumed rates of disassembly. Our experimental data were consistent with a disassembly rate per microtubule of 60 microns/min. This is the total rate of depolymerization for microtubules in the rapid shortening phase, as determined by light microscopy of individual microtubules (Walker et al.: Journal of Cell Biology 107:1437-1448, 1988). We conclude, therefore, that microtubules began rapid shortening at both ends upon dilution. Moreover, since we could detect no lag between dilution and the onset of rapid disassembly, the transition from elongation to rapid shortening apparently occurred within 1 sec following dilution. Assuming that this transition (catastrophe) involves the loss of the GTP cap, and that cap loss is achieved by the sequential dissociation of GTP-tubulin subunits following dilution, we can estimate the maximum size of the cap based on the kinetic data and model interpretation of Walker et al. The cap is probably shorter than 40 and 20 subunits at the plus and minus ends, respectively.     

Dynamics of interphase microtubules in Schizosaccharomyces pombe

BACKGROUND: Microtubules in interphase Schizosaccharomyces pombe are essential for maintaining the linear growth habit of these cells. The dynamics of assembly and disassembly of these microtubules are so far uncharacterised. RESULTS: Live cell confocal imaging of alpha1 tubulin tagged with enhanced green fluorescent protein revealed longitudinally oriented, dynamically unstable interphase microtubule assemblies (IMAs). The IMAs were uniformly bright along their length apart from a zone of approximately doubly intense fluorescence commonly present close to their centres. The ends of each IMA switched from growth ( approximately 3.0 microm/min) to shrinkage ( approximately 4.5 microm/min) at 1.0 events per minute and from shrinkage to growth at 1.9 events per minute, and the two ends were equivalently dynamic, suggesting equivalent structure. We accordingly propose a symmetrical model for microtubule packing within the IMAs, in which microtubules are plus ends out and overlap close to the equator of the cell. IMAs may contain multiple copies of this motif; if so, then within each IMA end, the microtubule ends must synchronise catastrophe and rescue. When both ends of an IMA lodge in the hemispherical cell ends, the IMAs start to bend under compression and their overall growth rate is inhibited about twofold. Similar microtubule dynamics were observed in cells ranging in size from half to twice normal length. Patterned photobleaching indicated no detectable treadmilling or microtubule sliding during interphase. CONCLUSIONS: The consequence of the mechanisms described is continuous recruitment of microtubule ends to the ends of growing cells, supporting microtubule-based transport into the cell ends and qualitatively accounting for the essential role for microtubules in directing linear cell growth in S. pombe.     

Membrane/microtubule tip attachment complexes (TACs) allow the assembly dynamics of plus ends to push and pull membranes into tubulovesicular networks in interphase Xenopus egg extracts

We discovered by using high resolution video microscopy, that membranes become attached selectively to the growing plus ends of microtubules by membrane/microtubule tip attachment complexes (TACs) in interphase- arrested, undiluted, Xenopus egg extracts. Persistent plus end growth of stationary microtubules pushed the membranes into thin tubules and dragged them through the cytoplasm at the approximately 20 microns/min velocity typical of free plus ends. Membrane tubules also remained attached to plus ends when they switched to the shortening phase of dynamic instability at velocities typical of free ends, 50-60 microns/min. Over time, the membrane tubules contacted and fused with one another along their lengths, forming a polygonal network much like the distribution of ER in cells. Several components of the membrane networks formed by TACs were identified as ER by immunofluorescent staining using antibodies to ER-resident proteins. TAC motility was not inhibited by known inhibitors of microtubule motor activity, including 5 mM AMP-PNP, 250 microM orthovanadate, and ATP depletion. These results show that membrane/microtubule TACs enable polymerizing ends to push and depolymerizing ends to pull membranes into thin tubular extensions and networks at fast velocities.     

Structural plugs at microtubule ends may regulate polymer dynamics in vitro

Microtubules contain in their lumens distinct structures (plugs) that influence their dynamic behavior in vitro. As observed by electron microscopy, plugs are stain-occluding structures 10-30 nm in length that occur along the lengths and at the ends of microtubules. Plugs occur at a frequency of 20-40% at the ends of microtubules assembled from cycled microtubule protein containing MAPs. While the composition of plugs is not known, preliminary evidence suggests that they are accretions of tubulin, that they are labile, and that they are more common in preparations containing MAPs. When polymers are induced to depolymerize by endwise subunit dissociation, the frequency of plugged microtubule ends increases transiently, suggesting that plugs temporarily stabilize microtubules. The functional significance of plugs may be that they prevent the sudden complete loss of microtubules through catastrophic disassembly. It is possible that plugs, by slowing the rate of disassembly, enable a polymer to add GTP-tubulin subunits, thereby forming a stabilizing GTP-cap. These observations suggest that plugs may stabilize polymers and account for the frequent transitions from shortening to growing phases that characterize dynamic instability.     

Dynamic Behavior of Microtubules during Dynein-dependent Nuclear Migrations of Meiotic Prophase in Fission Yeast

During meiotic prophase in fission yeast, the nucleus migrates back and forth between the two ends of the cell, led by the spindle pole body (SPB). This nuclear oscillation is dependent on astral microtubules radiating from the SPB and a microtubule motor, cytoplasmic dynein. Here we have examined the dynamic behavior of astral microtubules labeled with the green fluorescent protein during meiotic prophase with the use of optical sectioning microscopy. During nuclear migrations, the SPB mostly follows the microtubules that extend toward the cell cortex. SPB migrations start when these microtubules interact with the cortex and stop when they disappear, suggesting that these microtubules drive nuclear migrations. The microtubules that are followed by the SPB often slide along the cortex and are shortened by disassembly at their ends proximal to the cortex. In dynein-mutant cells, where nuclear oscillations are absent, the SPB never migrates by following microtubules, and microtubule assembly/disassembly dynamics is significantly altered. Based on these observations, together with the frequent accumulation of dynein at a cortical site where the directing microtubules interact, we propose a model in which dynein drives nuclear oscillation by mediating cortical microtubule interactions and regulating the dynamics of microtubule disassembly at the cortex.     

Stathmin family protein SCG10 differentially regulates the plus and minus end dynamics of microtubules at steady state in vitro: implications for its role in neurite outgrowth

SCG10 (superior cervical ganglia neural-specific 10 protein) is a neuron specific member of the stathmin family of microtubule regulatory proteins that like stathmin can bind to soluble tubulin and depolymerize microtubules. The direct actions of SCG10 on microtubules themselves and on their dynamics have not been investigated previously. Here, we analyzed the effects of SCG10 on the dynamic instability behavior of microtubules in vitro, both at steady state and early during microtubule polymerization. In contrast to stathmin, whose major action on dynamics is to destabilize microtubules by increasing the switching frequency from growth to shortening (the catastrophe frequency) at microtubule ends, SCG10 stabilized the plus ends both at steady state and early during polymerization by increasing the rate and extent of growth. For example, early during polymerization at high initial tubulin concentrations (20 microM), a low molar ratio of SCG10 to tubulin of 1:30 increased the growth rate by approximately 50%. In contrast to its effects at plus ends, SCG10 destabilized minus ends by increasing the shortening rate, the length shortened during shortening events, and the catastrophe frequency. Consistent with its ability to modulate microtubule dynamics at steady state, SCG10 bound to purified microtubules along their lengths. The dual activity of SCG10 at opposite microtubule ends may be important for its role in regulating growth cone microtubule dynamics. SCG10's ability to promote plus end growth may facilitate microtubule extension into filopodia, and its ability to destabilize minus ends could provide soluble tubulin for net plus end elongation.     

The polarity and dynamics of microtubule assembly in the budding yeast Saccharomyces cerevisiae

Microtubule assembly in Saccharomyces cerevisiae is initiated from sites within spindle pole bodies (SPBs) in the nuclear envelope. Microtubule plus ends are thought to be organized distal to the SPBs, while minus ends are proximal. Several hypotheses for the function of microtubule motor proteins in force generation and regulation of microtubule assembly propose that assembly and disassembly occur at minus ends as well as at plus ends. Here we analyse microtubule assembly relative to the SPBs in haploid yeast cells expressing green fluorescent protein fused to alpha-tubulin, a microtubule subunit. Throughout the cell cycle, analysis of fluorescent speckle marks on cytoplasmic astral microtubules reveals that there is no detectable assembly or disassembly at minus ends. After laser-photobleaching, metaphase spindles recover about 63% of the bleached fluorescence, with a half-life of about 1 minute. After anaphase onset, photobleached marks in the interpolar spindle are persistent and do not move relative to the SPBs. In late anaphase, the elongated spindles disassemble at the microtubule plus ends. These results show for astral and anaphase interpolar spindle microtubules, and possibly for metaphase spindle microtubules, that microtubule assembly and disassembly occur at plus, and not minus, ends.     

Polymerization of tubulin in vivo: direct evidence for assembly onto microtubule ends and from centrosomes

Microtubule assembly in vivo was studied by hapten-mediated immunocytochemistry. Tubulin was derivatized with dichlorotriazinylaminofluorescein (DTAF) and microinjected into living, interphase mammalian cells. Sites of incorporation were determined at the level of individual microtubules by double-label immunofluorescence. The haptenized tubulin was localized by an anti-fluorescein antibody and a second antibody conjugated with fluorescein. Total microtubules were identified by anti-tubulin and a secondary antibody conjugated with rhodamine. Contrary to recent studies (Salmon, E. D., et al., 1984, J. Cell Biol., 99:2165-2174; Saxton, W. M., et al., 1984, J. Cell Biol., 99:2175-2186) which suggest that tubulin incorporates all along the length of microtubules in vivo, we found that microtubule assembly in interphase cells was in vivo, as in vitro, an end-mediated process. Microtubules that radiated out toward the cell periphery incorporated the DTAF-tubulin solely at their distal, that is, their plus ends. We also found that a proportion of the microtubules connected to the centrosomes incorporated the DTAF-tubulin along their entire length, which suggests that the centrosome can nucleate the formation of new microtubules.     

Phase dynamics at microtubule ends: the coexistence of microtubule length changes and treadmilling

The length dynamics both of microtubule-associated protein (MAP)-rich and MAP-depleted bovine brain microtubules were examined at polymer mass steady state. In both preparations, the microtubules exhibited length redistributions shortly after polymer mass steady state was attained. With time, however, both populations relaxed to a state in which no further changes in length distributions could be detected. Shearing the microtubules or diluting the microtubule suspensions transiently increased the extent to which microtubule length redistributions occurred, but again the microtubules relaxed to a state in which changes in the polymer length distributions were not detected. Under steady-state conditions of constant polymer mass and stable microtubule length distribution, both MAP-rich and MAP-depleted microtubules exhibited behavior consistent with treadmilling. MAPs strongly suppressed the magnitude of length redistributions and the steady-state treadmilling rates. These data indicate that the inherent tendency of microtubules in vitro is to relax to a steady state in which net changes in the microtubule length distributions are zero. If the basis of the observed length redistributions is the spontaneous loss and regain of GTP-tubulin ("GTP caps") at microtubule ends, then in order to account for stable length distributions the microtubule ends must reside in the capped state far longer than in the uncapped state, and uncapped microtubule ends must be rapidly recapped. The data suggest that microtubules in cells may have an inherent tendency to remain in the polymerized state, and that microtubule disassembly must be induced actively.     

Inhibition of neurite initiation and growth by taxol

We cultured sensory neurons from chick embryos in media containing the alkaloid taxol at concentrations from 7 X 10(-9) to 3.5 X 10(-6) M. When plated at taxol concentrations above 7 X 10(-8) M for 24 h, neurons have short broad extensions that do not elongate on the culture substratum. When actively growing neurites are exposed to these levels of taxol, neurite growth stops immediately and does not recommence. The broad processes of neurons cultured 24 h with taxol contain densely packed arrays of microtubules that loop back at the ends of the process. Neurofilaments are segregated from microtubules into bundles and tangled masses in these taxol-treated neurons. At the ends of neurites treated for 5 min with taxol, microtubules also turn and loop back abnormally toward the perikaryon. In the presence of 7 X 10(-9) M taxol neurites do grow, although they are broader and less branched than normally. The neurites of these cells appear to have normal structure except for a large number of microtubules. Taxol probably stimulates microtubule polymerization in these cultured neurons. At high levels of the drug, this action inhibits neurite initiation and outgrowth by removing free tubulin from the cytoplasm and destroying the normal control of microtubule assembly in growing neurites. The rapid inhibition suggests that microtubule assembly may occur at neurite tips. At lower concentrations, taxol may slightly enhance the mechanisms of microtubule assembly in neurons, and this alteration of normal processes changes the morphogenetic properties of the growing neurites.     

Spindle microtubule differentiation and deployment during micronuclear mitosis in Paramecium

Spindles underwent a 12-fold elongation before anaphase B was completed during the closed mitoses of micronuclei in Paramecium tetraurelia. Two main classes of spindle microtubules have been identified. A peripheral sheath of microtubules with diameters of 27-32 nm was found to be associated with the nuclear envelope and confined to the midportion of each spindle. Most of the other microtubules had diameters of approximately 24 nm and were present along the entire lengths of spindles. Nearly all of the 24-nm microtubules were eliminated from spindle midportions (largely because of microtubule disassembly) at a relatively early stage of spindle elongation. Disassembly of some of these microtubules also occurred at the ends of spindles. About 60% of the total microtubule content of spindles was lost at this stage. Most, perhaps all, peripheral sheath microtubules remained intact. Many of them detached from the nuclear envelope and regrouped to form a compact microtubule bundle in the spindle midportion. There was little, if any, further polymerization of 24-nm microtubules after the disassembly phase. Polymerization of microtubules with diameters of 27-32 nm continued as spindle elongation progressed. Most microtubules in the midportions of well-elongated spindles were constructed from 14-16 protofilaments. A few 24-nm microtubules with 13 protofilaments were also present. The implications of these findings for spatial control of microtubule assembly, disassembly, positioning, and membrane association, that apparently discriminate between microtubules with different protofilament numbers have been explored. The possibility that microtubule sliding occurs during spindle elongation has also been considered.     

COMPARISON OF THE BEHAVIOUR OF A SOLUBLE AND A MEMBRANE-BOUND ENZYME IN TRANSECTED PERIPHERAL NERVES

Phosphoglucoisomerase (PGI), a soluble enzyme, and AChE, a membrane-bound enzyme were studied in transected peroneal nerves of dog and in isolated segments of these nerves. Although activities of both enzymes increased at the ends of transected nerves, marked differences in their behaviour were observed. The increment in AChE activity was much sharper than that of PGI and continued to grow with time whereas the increase in PGI developed fully within the initial hours after transection and did not change thereafter. In an isolated nerve segment AChE accumulated at both ends with a concomitant decrease in the middle part, whereas changes in PGI activity appeared only in the terminal parts, the rest of the nerve remaining at the normal level. The terminal increase of PGI did not, contrary to that of AChE, depend on the length of the isolated segment. The changes in PGI activity may be features of a local peritraumatic reaction whereas those of AChE indicate involvement of the whole segment along which the enzyme containing organelles are transported.     

Optimizing microtubule arrangements for rapid cargo capture

Cellular functions such as autophagy, cell signaling, and vesicular trafficking involve the retrograde transport of motor-driven cargo along microtubules. Typically, newly formed cargo engages in slow undirected movement from its point of origin before attaching to a microtubule. In some cell types, cargo destined for delivery to the perinuclear region relies on capture at dynein-enriched loading zones located near microtubule plus ends. Such systems include extended cell regions of neurites and fungal hyphae, where the efficiency of the initial diffusive loading process depends on the axial distribution of microtubule plus ends relative to the initial cargo position. We use analytic mean first-passage time calculations and numerical simulations to model diffusive capture processes in tubular cells, exploring how the spatial arrangement of microtubule plus ends affects the efficiency of retrograde cargo transport. Our model delineates the key features of optimal microtubule arrangements that minimize mean cargo capture times. Namely, we show that configurations with a single microtubule plus end abutting the distal tip and broadly distributed other plus ends allow for efficient capture in a variety of different scenarios for retrograde transport. Live-cell imaging of microtubule plus ends in Aspergillus nidulans hyphae indicates that their distributions exhibit these optimal qualitative features. Our results highlight important coupling effects between the distribution of microtubule tips and retrograde cargo transport, providing guiding principles for the spatial arrangement of microtubules within tubular cell regions.     

Polarity of microtubules nucleated by centrosomes and chromosomes of Chinese hamster ovary cells in vitro

The structural and growth polarities of centrosomal and chromosomal microtubules were studied by analyzing the kinetics of growth of these microtubules and those initiated by flagellar seeds. By comparing rates of elongation of centrosomal and flagellar-seeded microtubules, we determined whether the centrosomal microtubules were free to grow at their plus ends only, minus ends ony, or at both ends. Our results show that centrosomal microtubules elongate at a rate corresponding to the addition of subunits at the plus end only. The depolymerization rate was also equivalent to that for the plus end only. Chromosomal microtubule elongation was similar to the centrosome-initiated growth. Since the data do not support the hypothesis that both ends of these spindle microtubules are able to interact with monomer in solution, then growth must occur only distal or only proximal to the organizing centers, implying tha the opposite ends in unavailable for exchange of subunits. Experiments with flagellar-seeded microtubules serving as internal controls indicated that the inactivity of the minus end could not be accounted for by a diffusible inhibitor, suggesting a structural explanation. Since there is no apparent way in which the distal ends may be capped, whereas the proximal ends are embedded in the pericentriolar cloud, we conclude that centrosomal microtubules are oriented with their plus ends distal to the site of nucleation. A similar analysis for chromosomal microtubules suggests that they too must be oriented with their plus ends distal to the site of initiation.     

The Arabidopsis CLASP gene encodes a microtubule-associated protein involved in cell expansion and division

Controlling microtubule dynamics and spatial organization is a fundamental requirement of eukaryotic cell function. Members of the ORBIT/MAST/CLASP family of microtubule-associated proteins associate with the plus ends of microtubules, where they promote the addition of tubulin subunits into attached kinetochore fibers during mitosis and stabilize microtubules in the vicinity of the plasma membrane during interphase. To date, nothing is known about their function in plants. Here, we show that the Arabidopsis thaliana CLASP protein is a microtubule-associated protein that is involved in both cell division and cell expansion. Green fluorescent protein-CLASP localizes along the full length of microtubules and shows enrichment at growing plus ends. Our analysis suggests that CLASP promotes microtubule stability. clasp-1 T-DNA insertion mutants are hypersensitive to microtubule-destabilizing drugs and exhibit more sparsely populated, yet well ordered, root cortical microtubule arrays. Overexpression of CLASP promotes microtubule bundles that are resistant to depolymerization with oryzalin. Furthermore, clasp-1 mutants have aberrant microtubule preprophase bands, mitotic spindles, and phragmoplasts, indicating a role for At CLASP in stabilizing mitotic arrays. clasp-1 plants are dwarf, have significantly reduced cell numbers in the root division zone, and have defects in directional cell expansion. We discuss possible mechanisms of CLASP function in higher plants.     

The polarity and stability of microtubule capture by the kinetochore

We have studied the capture of microtubules by isolated metaphase chromosomes, using microtubules stabilized with taxol and marked with biotin tubulin to distinguish their plus and minus ends. The capture reaction is reversible at both the plus and minus ends. The on rate of capture is the same for both polarities but the dissociation rate from the kinetochore is seven times slower with microtubules captured at their plus ends than those captured at their minus ends. At steady state this disparity in off rates leads to the gradual replacement of microtubules captured at their minus ends with those captured at their plus ends. These results suggest that the kinetochore makes a lateral attachment near the end of the microtubule in the initial capture reaction and shows a structural specificity that may be important in proper bipolar attachment of the chromosome to the spindle.     

Suppression of microtubule dynamic instability by the +TIP protein EB1 and its modulation by the CAP-Gly domain of p150glued

The EB1+TIP protein family and its binding partners track growing plus ends of microtubules in cells and are thought to regulate their dynamics. Here we determined the effects of EB1 and the N-terminal CAP-Gly domain (p150n) of one of its major binding partners, p150Glued, both separately and together, on the dynamic instability parameters at plus ends of purified steady-state microtubules. With EB1 alone, the shortening rate, the extent of shortening, and the catastrophe frequency were suppressed in the absence of significant effects on the growth rate or rescue frequency. The effects of EB1 on dynamics were significantly different when p150n was added together with EB1. The rate and extent of shortening and the catastrophe frequency were suppressed 3-4 times more strongly than with EB1 alone. In addition, the EB1-p150n complex increased the rescue frequency and the mean length the microtubules grew, parameters that were not significantly affected by EB1 alone. Similarly, deletion of EB1's C-terminal tail, which is a crucial binding region for p150n, significantly increased the ability of EB1 to suppress shortening dynamics. EB1 by itself bound along the length of the microtubules with 1 mol of EB1 dimer bound per approximately 12 mol of tubulin dimer. Approximately twice the amount of EB1 was recruited to the microtubules in the presence of p150n. Our results indicate that inactivation of EB1's flexible C-terminal tail significantly changes EB1's ability to modulate microtubule dynamics. They further suggest that p150Glued may activate and thereby facilitate the recruitment of EB1 to the tips of microtubules to regulate their dynamics.     

Structure of growing microtubule ends: two-dimensional sheets close into tubes at variable rates

Observation of microtubule growth at different rates by cryo-electron microscopy reveals that the ends range from blunt to long, gently curved sheets. The mean sheet length increases with the growth rate while the width of the distributions increases with the extent of assembly. The combination of a concentration dependent growth rate of the tubulin sheet with a variable closure rate of the microtubule cylinder, results in a model in which stochastic fluctuations in sheet length and tubulin conformation confine GTP-tubulins to microtubule ends. We propose that the variability of microtubule growth rate observed by video microscopy (Gildersleeve, R. F., A. R. Cross, K. E. Cullen, A. P. Fagen, and R. C. Williams. 1992. J. Biol. Chem. 267: 7995- 8006, and this study) is due to the variation in the rate of cylinder closure. The curvature of the sheets at the end of growing microtubules and the small oligomeric structures observed at the end of disassembling microtubules, indicate that tubulin molecules undergo conformational changes both during assembly and disassembly.