Research Article

Sustained Microtubule Treadmilling in Arabidopsis Cortical Arrays

See allHide authors and affiliations

Science  13 Jun 2003:
Vol. 300, Issue 5626, pp. 1715-1718
DOI: 10.1126/science.1083529


Plant cells create highly structured microtubule arrays at the cell cortex without a central organizing center to anchor the microtubule ends. In vivo imaging of individual microtubules in Arabidopsis plants revealed that new microtubules are initiated at the cell cortex and exhibit dynamics at both ends. Polymerization-biased dynamic instability at one end and slow depolymerization at the other end result in sustained microtubule migration across the cell cortex by a hybrid treadmilling mechanism. This motility causes widespread microtubule repositioning and contributes to changes in array organization through microtubule reorientation and bundling.

The microtubule cytoskeleton plays a key role in plant cell morphogenesis and multicellular development. Disruption of plant microtubule organization by drugs or through mutation causes defects ranging from changes in cell shape to a marked loss of organ form (15). The cortical microtubule array is proposed to influence cell shape by guiding the deposition of new cell wall polymers (4, 6, 7). Aligning the cellulose microfibrils in the cell wall restricts cell elongation, resulting in anisotropic cell wall expansion and the acquisition of specialized cell shapes (7). How the cortical microtubules are created and positioned to form organized arrays is not known.

Plant cortical microtubule arrays are dynamic structures (8) that reorganize in response to environmental and developmental cues (2, 9). In epidermal cells of the root or shoot axis, interphase microtubules show a progressive change in organization after cell division. Unlike interphase arrays in animal or fungal cells, the plant cortical array does not radiate from a central organizer. Microtubules first appear at the cell cortex in a disordered arrangement and then form remarkable transverse helical arrays that change pitch as the cell expands (6, 9, 10).

Several mechanisms have been proposed for the creation and dynamic organization of the cortical arrays (2, 1115), including de novo microtubule polymerization in a preferred orientation, transport of microtubules originating at the nuclear surface to defined cortical positions, lateral and axial sliding (translocation) of existing cortical microtubules into new positions, and migration of polymers to new positions by the balanced addition and removal of subunits at the microtubule ends (treadmilling). To analyze the behavior of individual cortical microtubules and to explore which of these mechanisms contribute to cortical array organization, we created green fluorescent protein (GFP)–tubulin fusions that permit imaging of individual microtubules in transgenic Arabidopsis plants.

Individual microtubule behaviors. Time-lapse confocal imaging of Arabidopsis epidermal cells expressing GFP-tubulin fusion proteins (n = 33 cells, average duration 4.1 min at intervals of 3.85 s) revealed discrete sites of apparent microtubule initiation at the cell cortex (Fig. 1, A and B). Microtubules that depolymerized to visible completion rarely showed recovery (48 of 50 events with >2 min of imaging after depolymerization), which suggests that most initiation sites represent de novo origins. Initiation sites were scattered throughout the observed area of the cell cortex and appeared both in association with existing microtubules and in regions with no other detectable microtubules. In several cases, multiple microtubules polymerized from the same site (Fig. 1A) (movies S1 and S2). Limited examples of severing or breakage in elongated microtubules were recorded, although in 24 of 51 observed severing events at least one of the resulting microtubules depolymerized to extinction. In observations of 30 cells, we failed to observe a microtubule emerging from the cytosol to join the cortical array, whereas we observed 71 cortical initiation events in these same cells. Thus, the majority of new microtubules in mature interphase arrays were likely created at the cell cortex and did not come from interior organizing centers such as the nuclear surface. Consistent with this finding, AtSpc98, a proposed microtubule organizing center component, has been localized to the plasma membrane of Arabidopsis cells (16).

Fig. 1.

Microtubule initiation and unidirectional motility. (A) Time series (left to right) of two new microtubules (solid and open arrowheads) polymerizing from a site at the cell cortex (arrow) and diverging from this origin at different angles. (B) A newly polymerized microtubule (solid arrowhead) detaching from a cortical site of origin (arrow). After detachment, a second microtubule (open arrowhead) is initiated at the same location. (C) Motile microtubule (solid arrowhead) crossing one microtubule (open arrowhead) before encountering a second polymer and bundling (arrows). (D to F) Kymographs showing three single microtubules from the same cell, digitally linearized and moving from left to right. Single microtubules show dynamic instability at the leading end and primarily slow shortening at the lagging end. Note that the microtubule in (D) depolymerizes to extinction from the leading end at 170 s. Scale bar, 2.5 μm; images selected from time series of duration 112 s (A), 162 s (B), and 142 s (C).

New microtubules did not remain anchored to their site of initiation (Fig. 1A) (movies S1 to S3). Initiating microtubules extended several micrometers before either shortening to extinction (7 of 43 initiation events) or moving away from the initiation site (36 of 43). Motility was unidirectional, with single microtubules moving most commonly in shallow arcs (Fig. 1, B and C) (movies S1 to S3), often changing trajectory several times on the cell cortex during the course of observation (2 to 8 min). In cases where multiple microtubules initiated from the same location, they often departed at diverging angles (Fig. 1, A and B) (movies S1 to S3). Time-lapse images of digitally linearized microtubules further illustrated the unidirectional motility and revealed markedly different dynamic properties for the leading and lagging ends (Fig. 1, D to F, and below). After moving away from their sites of initiation, a subset of polymers shortened completely from the leading end, and the remainder migrated across the cell cortex and gradually elongated or became incorporated into microtubule bundles (Fig. 1, D to F).

To test whether the microtubule motility was due to sliding (translocation of the polymer) or treadmilling, we marked the GFP-labeled microtubules by photobleaching (n = 12 cells, >50 single microtubules). Although microtubules remained motile after photobleaching, the photobleached marks maintained fixed positions with reference to the cell (Fig. 2A) (movie S4); in no case did we observe the movement of a photobleached mark on a single microtubule (Fig. 2, B to D). Thus, single microtubules were fixed in place at the cell cortex, and the apparent microtubule motility resulted from polymerization and depolymerization at the ends and not from translocation of the intact microtubule polymer.

Fig. 2.

Microtubules associate with the cortex and move by treadmilling. (A) Marks generated by photobleaching GFP-tubulin–labeled microtubules do not move; hence, microtubule motility does not occur by translocation of existing polymer. Scale bar, 5 μm. Time before and after bleaching is given in seconds. (B to D) Kymographs representing polymers from the entire time sequence in (A) show that bleach marks remain fixed in place. (B) The bleached zone does not spread in microtubule bundles, and the bleach border does not travel into the bleached zone. (C) Single microtubule migrating from left to right with fixed bleach zone. (D) Bundle showing fluorescence recovery via constituent microtubules. Vertical bar, 2 min; horizontal bar, 2.5 μm. (E) First image in a time-lapse series (3.8-s intervals) of a cortical microtubule array. (F) Kymograph of a linear transect across this cell [gray line in (E)] at each time interval. Straight lines in the kymograph indicate no lateral movement of microtubules during the 6.5-min experiment. Vertical bar, 2.5 min (F); horizontal bar, 5 μm (E and F). (G) Detachment from the cell cortex is observed as rapid, lateral, and out-of-focus movement of microtubule ends (arrows), resulting in reattachment and reorientation (arrowheads) or depolymerization. Total time interval for image series, 96 s; scale bar, 2.5 μm.

Motile microtubules were observed either to cross over other microtubules or to incorporate into bundles (Fig. 1C) (movies S1 to S3). Bundling initiated when the leading end of a motile microtubule contacted another microtubule or bundle and changed trajectory to become co-aligned. Progressive changes in fluorescence intensity along the encountered polymer suggest that the leading end continued to polymerize along the bundle after initial contact. Depolymerization from the lagging end completed the bundling process by consuming the unbundled portion of the microtubule. Photobleached marks made on microtubule bundles typically recovered rapidly but did not move (Fig. 2, A, B, and D) (movie S4). Thus, bundled microtubules remained dynamic, with the observed dynamic behavior being caused by polymerization and depolymerization and not by the sliding of bundled microtubules.

To investigate whether lateral movements contributed to microtubule positioning, we drew a linear transect across the cell image to sample random discrete locations along multiple microtubules (Fig. 2E). The fluorescence signal along this transect from a time-lapse series of images was projected as a kymograph to analyze the stability of microtubule position over time (Fig. 2F). Parallel, vertical lines in the kymograph indicate that the cortical microtubules showed almost no lateral translocation, despite rapid cytoplasmic streaming. This lateral stability was evident even when single microtubules displayed no overlap with other cortical microtubules; hence, stabilization of microtubule position did not rely on intermicrotubule cross-linking. The stability of the fluorescence signal in the focal plane over the duration of the experiment further indicated that microtubules did not show detectable movement on and off the cell cortex. Thus, the majority of the microtubules in the observed cortical arrays were strongly associated with the cell cortex, as proposed previously from ultrastructural studies (17), and lateral translocation of both single and bundled microtubules was either rare or too slow to detect over a 6-min observation interval.

Exceptions to cortical association were found when a microtubule end moved rapidly out of focus and into the streaming cytoplasm (Fig. 2G) (movies S1 to S3). The abrupt loss of cortical association occurred at the leading end of the motile microtubule (69 of 71 events, n = 33 cells) and was observed almost exclusively for single microtubules (70 of 71 events), not for microtubules in bundles. The detachment of a free end resulted in complete depolymerization in slightly more than half of the cases (37 of 71 events); in the other 34 events, reassociation with the cortex occurred, often reorienting the microtubule. Thus, cortical attachment may be important for array organization because loss of attachment has consequences for polymer stability and orientation. Also, bundling might protect microtubules from cortical detachment, possibly through intramicrotubule cross-linking by other proteins (18).

Polymerization dynamics. To investigate how polymerization dynamics contribute to cortical array behavior, we measured the dynamic properties of both the cortical array and the individual microtubule ends. Using fluorescence redistribution after photobleaching (FRAP), we measured a recovery half-time of 58.95 s (n = 27 cells, SD = 14.7, SEM = 2.83) from epidermal cells in the hypocotyl (Fig. 3, A and B) (movie S5). This recovery time is faster than that measured in animal interphase arrays by a factor of ∼4 (19, 20). These data obtained with a GFP-tubulin fusion protein expressed in Arabidopsis confirm previous FRAP results from Tradescantia stamen hair cells injected with fluorescent animal tubulin (8).

Fig. 3.

Microtubule dynamics measurements. (A) Cortical microtubule array in a hypocotyl cell immediately after laser-mediated photobleaching of a 10-μm circle. Scale bar, 10 μm. (B) Fluorescence redistribution after photobleaching. Two images were taken before photobleaching and the remainder at 8- to 12-s intervals after bleaching. Fluorescence recovery measurements (mean ± SD, n = 27 cells) are corrected for photobleaching and normalized to the initial fluorescence value before fitting with an exponential function (solid line) (8). The curve is fit without the first measurement after photobleaching (red symbols) to correct for bleaching of unincorporated tubulin dimer. (C) Growth and shortening velocities for the leading and lagging ends of single microtubules recorded from 18 Arabidopsis epidermal cells. Measurements consist of 3064 velocities per end, representing 3.5 hours total time (∼3.8-s intervals, 78 microtubules). The histogram is color-coded for growth (green), pause (blue), and shortening (red) velocities. Mean velocities for growth or shortening are depicted next to the histograms.

To determine how individual microtubule polymerization dynamics contribute to cortical array turnover and behavior, we measured velocities of growth and shortening (Fig. 3C) and transition frequencies between growth, shortening, and pause states for single microtubules where both ends were clearly visible (Table 1, n = 78 microtubules from 18 cells). The microtubule end leading the unidirectional motility displayed five times as much net polymerization-depolymerization per unit time (dynamicity) as the lagging end (Table 1); this finding confirmed that the two ends had distinct dynamic properties. The leading end showed persistent phases of both growth and shortening, with the rate of shortening faster on average (5.88 ± 5.07 μm/min) than the rate of growth (3.69 ± 1.90 μm/min). Catastrophe (0.043 s1) and rescue (0.082 s1) frequencies, however, favored time spent in growth (Table 1), resulting in a net gain in polymer at the leading end. Lagging end growth was slow (1.96 ± 1.24 μm/min) and rare (Table 1), possibly falling within the error of the measurement technique. The lagging end spent approximately the same amount of time shortening as did the leading end, but depolymerization occurred more slowly (2.78 ± 2.13 μm/min) and commonly transitioned to a pause state (rescue = 0.128 s1, Table 1). We measured an increase in total polymer for the microtubules sampled in this study (0.36 μm/min per microtubule). This increase arose at least in part because microtubules that elongated and then associated into bundles could no longer be measured (Fig. 1C), but may also reflect an actual bias in the dynamics of the population of sampled microtubules.

Table 1.

In vivo transition rates for single microtubules. K, rate of transitions between dynamic states, measured at 3.8- to 5-s intervals; g, growth; s, shorten; p, pause.

Microtubule end
Lagging Leading
Transitions (events per minute)
    Kg-g 0.21 6.75
    Kg-s 0.52 0.97
    Kg-p 0.28 0.47
    Kp-g 0.26 0.51
    Kp-p 6.52 0.12
    Kp-s 1.30 0.24
    Ks-p 1.09 0.23
    Ks-g 0.59 0.87
    Ks-s 1.40 2.03
Events per second (in phase)
    Rescue 0.128 0.082
    Catastrophe 0.190 0.043
Time in phase
    Growth 8.4% 65.3%
    Pause 66.3% 10.1%
    Shorten 25.3% 24.6%
Dynamicity (μm/min)
0.83 ± 0.74 4.10 ± 1.41

Conclusions. Microtubules were observed to migrate across the cortex of Arabidopsis epidermal cells by means of a hybrid treadmilling mechanism. Treadmilling motility was not caused by pronounced dynamic instability at both polymer ends (21), nor by steady gain at one end and steady loss at the other (22). Rather, motility was the net result of slow, intermittent depolymerization at the lagging end, coupled with polymerization-biased dynamic instability at the leading end. Further, the discovery of dynamic instability as the dominant mode of dynamic behavior in the Arabidopsis interphase arrays suggests that dynamic instability is integral to the organization of both centriolar and acentriolar interphase arrays.

Treadmilling events have been observed in animal cells and cytoplasts when microtubules escaped from the centrosome or suffered breakage events (2328). These events are relatively rare and short-lived, ending by rapid depolymerization from the minus end (24) or by depolymerization from the plus end when the minus end is stabilized (25, 26). By contrast, treadmilling motility in Arabidopsis microtubules is neither rare nor short-lived. The majority of microtubules we measured (50 of 78) showed strictly defined treadmilling for 22.5% of the observation interval. The lagging end of the microtubule seldom remained stable over time (6 of 78 microtubules), and complete depolymerization of microtubules was observed to occur only from the leading end (n = 50 of 50 from 10 cells), even in cases of severing. The slow and intermittent depolymerization at the lagging end of plant interphase microtubules suggests that sustained treadmilling motility results from careful regulation of the minus end.

The extent of the treadmilling, the creation of microtubule bundles through treadmilling motility, and the absence of other observed mechanisms for polymer repositioning together suggest that treadmilling motility makes an important contribution to the organization of the cortical array.

Supporting Online Material

SOM Text

Materials and Methods


Movies S1 to S5

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article