Research Article

A Mechanism for Reorientation of Cortical Microtubule Arrays Driven by Microtubule Severing

See allHide authors and affiliations

Science  06 Dec 2013:
Vol. 342, Issue 6163, 1245533
DOI: 10.1126/science.1245533

Structured Abstract


Organization of the cortical cytoskeleton guides the growth and morphogenesis of organisms, from bacteria to higher plants, that depend on cell walls. By positioning wall-building enzymes, the cytoskeleton acts as an interior scaffold to direct construction of the cell’s exterior. In plants, environmental and hormonal signals that modulate cell growth cause reorganization of cortical microtubule arrays. These arrays do not appear to be remodeled by moving individual microtubules, but rather by rules that govern how microtubules are assembled or disassembled. In this Research Article, we investigate the mechanism by which blue light, an important signal from the environment, causes a rapid 90° reorientation of cortical arrays in growing cells of the plant axis.

Embedded Image

Blue light perception stimulates generation of a cascade of newly oriented microtubules by katanin severing. A confocal microscopy time series of the cortical microtubule array (white, preexisting; blue, newly assembled) in an Arabidopsis epidermal cell is shown. Perception of blue light by phototropin receptors has stimulated severing at microtubule intersections. Growth of the new ends creates new and co-oriented microtubules. Together, the organization of the preexisting array and the statistical behavior of severing favor the growth of longitudinal microtubules, driving array reorientation.


We used spinning-disk confocal microscopy to image the reorganization of cortical microtubule arrays in real time and visualize functional proteins tagged with fluorescent proteins. We developed image-analysis methods to measure changes in array organization and behaviors of individual microtubules during array reorientation. To test hypotheses about signaling and reorganizational mechanisms, we analyzed mutants in light-perception pathways and in activity of the microtubule-severing protein katanin. Finally, we conducted photomorphogenesis assays in plant seedlings to place our findings in a physiological context.


We discovered a mechanism, based on microtubule severing by the protein katanin, that reorients cortical microtubule arrays in response to perception of blue light. Specifically, we observed that katanin localized to microtubule crossovers, where it was required to preferentially catalyze the severing of the newer microtubule, an activity that was stimulated by the function of phototropin blue light receptors. New plus ends created by severing were stabilized and immediately grew at a high frequency, resulting in the effective creation of new microtubules. Most microtubules generated during reorientation were created by this mechanism, producing ~83% of new longitudinal microtubules. Cortical arrays failed to reorient in a mutant lacking the katanin protein. Microtubules produced by severing at crossovers can make new crossovers and, thus, opportunities for further rounds of severing and initiation, constituting a molecular amplifier that rapidly builds a new population of microtubules orthogonal to the initial array. Further experiments put this mechanism in a physiological context by revealing that katanin function is required for directional blue light to stimulate bending of the plant axis toward the light source.


Cortical microtubule arrays in higher plants are being recognized as systems with self-organizing properties arising from rules governing the outcomes of microtubule interactions. In this Research Article, we present evidence that one outcome of microtubule interaction, katanin-mediated severing at crossover sites, is regulated by light perception and acts to reorient the array. Severing is thought to help build microtubule arrays in neurons and meiocytes, but it has been difficult to test this idea directly because of imaging limitations. With live imaging of plant cell cortical arrays, we have been able to investigate the cellular function of severing at the level of individual molecular events, revealing how generation of microtubules by severing is used to construct a new array.

Light Turns the Array

The organization of cortical microtubule arrays in higher plant cells is essential for organizing cell and tissue morphogenesis, but it is not clear how specific architectures are acquired and reconfigured in response to environmental cues. Lindeboom et al. (10.1126/science.1245533, published online 7 November; see the Perspective by Roll-Mecak) used live-cell imaging and genetic studies to show that the microtubule-severing protein, katanin, plays a crucial role in reorienting cortical arrays from transverse to longitudinal in Arabidopsis seedlings in response to blue light perception. Katanin localized to microtubule intersections where, stimulated by blue light receptors, it preferentially catalyzed the severing of the newer microtubule. The microtubule “plus” end created by severing were observed to grow preferentially, effectively building a new population of microtubules orthogonal to the initial array. The net effect of this process steers the growing seedling toward light.


Environmental and hormonal signals cause reorganization of microtubule arrays in higher plants, but the mechanisms driving these transitions have remained elusive. The organization of these arrays is required to direct morphogenesis. We discovered that microtubule severing by the protein katanin plays a crucial and unexpected role in the reorientation of cortical arrays, as triggered by blue light. Imaging and genetic experiments revealed that phototropin photoreceptors stimulate katanin-mediated severing specifically at microtubule intersections, leading to the generation of new microtubules at these locations. We show how this activity serves as the basis for a mechanism that amplifies microtubules orthogonal to the initial array, thereby driving array reorientation. Our observations show how severing is used constructively to build a new microtubule array.

Cytoskeletal arrays based on tubulin are ancient and found across all cellular life. These arrays are organized into specific configurations that are necessary for diverse essential processes such as chromosome segregation, intracellular transport, cell motility, and generation of cell shape. To support these diverse functions, microtubule arrays are remodeled into new arrangements over the course of the cell cycle and in response to environmental and developmental information. In many cases, centralized bodies such as centrosomes act to organize microtubule arrays by tethering and positioning nucleation complexes, setting up their functional architecture. Though centrosomal arrays have been well studied, comparatively little is understood about how microtubule arrays are organized and dynamically reorganized without the aid of centrosomes, despite the fact that acentrosomal arrays are common across diverse phyla and are featured in most differentiated animal cells (1).

Movie S1

Watch Video

Lacking centrosomes altogether, higher plants create highly ordered microtubule arrays at the cell cortex. By organizing cell wall biogenesis and growth (24), these arrays direct cell and tissue morphogenesis and, thus, support essential functions such as photosynthesis, nutrient acquisition, and reproduction. The architectures of cortical microtubule arrays are dynamic and responsive to a range of signals (5), including light (6). In rapidly elongating cells of the embryonic shoot axis (a tissue known as the hypocotyl), cortical arrays are organized transversely to the axis of cell and tissue growth (Fig. 1A). Blue light stimulation, a potent signal to hypocotyl growth, causes these transverse arrays to undergo a 90° reorientation (79) within minutes (Fig. 1B), redirecting cellulose deposition to build and organize the cell wall (3). The molecular and cellular mechanisms driving reorientation of these acentrosomal arrays are not known. We investigated this question employing quantitative live-cell imaging and genetic tests, revealing a mechanism for reorientation that features the microtubule-severing protein katanin in creating microtubules to build a new array.

Fig. 1 Microtubule reorientation induced by blue light.

(A) Schematic drawing indicates location of imaging and angular coordinates. (B) Microtubule reorientation in an etiolated hypocotyl cell expressing mCherry-TUA5 and GCP2-3×GFP (only mCherry-TUA5 signal is shown). Scale bar, 5 µm. See also movie S1.

New Microtubules Primarily Initiate at Microtubule Crossovers During Reorientation

Movie S2

Watch Video

In the absence of a central organizer, the question of where and how new microtubules arise is critical for understanding how arrays are built and remodeled. To address this question, we imaged microtubule dynamics in elongating hypocotyl cells as they were stimulated by blue light delivered by the imaging energy itself. The location and orientation of all observable microtubule initiations in the imaged area of each cell were measured over the course of reorientation (Fig. 1B). To aid in the detection of new growing ends, we performed a walking image subtraction on the microtubule signal and selected the positive difference, a manipulation that results in clearly discerned “comets” (see materials and methods and movie S1). By revealing growing ends together with the labeled microtubule lattice, microtubule initiations could be distinguished from rescue events. During reorientation, we observed two distinct modes for the initiation of new microtubule growth, both of which were associated with existing microtubules. First, we observed events marked by a labeled γ-tubulin complex protein (GCP2-3×GFP; GCP2, γ-tubulin complex protein 2; GFP, green fluorescent protein) at the site of initiation, indicating the presence and likely participation of a γ-tubulin nucleation complex (Fig. 2A and movie S2) (10). Second, we observed frequent initiation events in which GCP2-3×GFP was not detected (Fig. 2, B and C, and movie S3). Almost all (98%) of these events arose from locations where one microtubule crossed over another. By contrast, nearly all (97%) microtubules marked by GCP2-3×GFP arose at locations other than crossovers (Fig. 2D; P < 0.00001, Fisher’s exact test). The initiations at crossovers were not rare, but in fact constituted the majority of all detected initiations (62%, 133 out of 212 initiations, seven plants). Thus, we discovered evidence for an atypical initiation mechanism acting at microtubule crossovers, which dominated during the course of array reorientation.

Fig. 2 Generation of new microtubules during light-induced microtubule reorientation.

(A) Example of a new microtubule (arrows) being generated from a GCP2-3×GFP–labeled complex (solid arrowheads). See also movie S2. (B) New microtubule growth (arrows) initiated at a microtubule crossover without GCP2-3×GFP signal (open arrowheads). See also movie S3. (C) Kymograph of the initiation shown in (B). Dashed blue line in top panel shows location of the kymograph line. Arrowheads show the position of microtubule initiation. (D) Fraction of new microtubules marked by a GCP2-3×GFP–labeled complex, classified by localization at microtubule crossovers or not at crossovers (N = 345 events in seven plants). (E) Relative frequencies of observed microtubule-severing events at crossovers in WT (N = 1115 events in five cells) and ktn1-2 mutant plants (N = 1304 in five cells). (F) Localization of GFP-KTN1 (yellow arrowheads) at an initiation event (blue arrowheads) from a crossover. See also movie S4. (G) Kymographs of microtubule initiation from a crossover from the dashed blue line in (F). Scale bars, 5 μm.

Katanin Localizes to Crossovers and Is Required to Initiate New Microtubules

Movie S3

Watch Video

Movie S4

Watch Video

Movie S5

Watch Video

Movie S6

Watch Video

Movie S7

Watch Video

Movie S8

Watch Video

We hypothesized that the atypical initiations at crossovers might be the result of severing followed by instantaneous growth of the newly created plus end. In experiments in which microtubules are severed in vitro, the newly created plus ends transition immediately to depolymerization (11, 12). Similar depolymerization had been observed to originate from microtubule crossovers in plant cell cortical arrays, implicating crossovers as sites of severing (13). We reasoned that if a population of plus ends produced by severing at crossovers immediately resumed growth, these would appear as new growing ends emerging from crossovers (fig. S1). If true, then disruption of the responsible severing protein would eliminate both formation of depolymerizing plus ends and atypical initiations at crossovers. A prime candidate for the severing protein was katanin (KTN1), which had been shown to sever microtubules in vitro (14). We assayed events at microtubule crossovers in ktn1-2, a mutant for the single gene in Arabidopsis encoding the catalytic katanin P60 subunit (10, 15). Whereas in wild-type (WT) plants, either creation of a new depolymerizing plus end (fig. S3A) or initiation of a new growing end (Fig. 2, B and F) was observed at 47% of the crossovers formed during the observation interval (522 out of 1115 crossovers), both outcomes were abolished in the mutant (0 out of 1304 crossovers) (Fig. 2E). Newly created minus ends were typically observed to be stable or to show slow loss (fig. S3A). Evidence for rapid depolymerization from new minus ends was not observed in this data set. We conclude that katanin is necessary for microtubule severing at crossovers and is required to generate the atypical initiations we observed at these locations, which are most simply explained as the immediate growth of plus ends created by severing (fig. S1).

As a further test for whether katanin acts directly to sever microtubules at crossovers, we asked if the katanin protein itself is recruited to these locations at the time of apparent severing. We expressed a GFP-KTN1 fusion protein in the ktn1-2 mutant, which complemented the ktn1-2 mutant phenotype (fig. S2), and showed distinct localization to microtubule crossovers, as visualized with coexpressed mCherry-TUA5 (4) (68% of crossovers at any time, 610 out of 891 crossovers in five plants). Recruitment of GFP-KTN1 to crossovers preceded 100% of events in which either discontinuities or new growing ends were observed (N = 402 events in five plants), and recruitment was lost after a short time lag (Fig. 2, F and G, fig. S3, and movies S4 to S6). The frequency with which evidence of severing was observed was not significantly dependent on the crossover angle (fig. S4).

When a crossover is made, a preexisting microtubule is crossed by a new microtubule. We found that the frequency of severing was not evenly distributed but occurred 1.7 times more often at the new microtubule (330 at the new microtubule versus 192 at the old; P < 0.0001, binomial test). Together, these data indicate that KTN1 is recruited to microtubule crossovers, where it preferentially catalyzes severing of the newer microtubule, often resulting in a new growing plus end.

New Microtubules Initiated by Severing Account for Most New Longitudinal Microtubules

Mathematical modeling of microtubule severing has shown that if severing leads only to depolymerizating new plus ends, such as observed in vitro, it does not increase the number of microtubules (16). Therefore, the observation of plus-end growth immediately after severing in vivo is noteworthy, because only those events have the potential to make a positive contribution to building a new array, as they effectively act as nucleations. To investigate how both nucleation events associated with γ-tubulin complexes and katanin-dependent initiations may contribute to the evolution of organization after blue light stimulation, we measured the angles of all new growing ends with respect to both the existing microtubules they were associated with and to cellular coordinates (Fig. 3, A to G). We further scored new growing ends for their association with GCP2-3×GFP. Consistent with previous observations (10, 17), most new growing ends marked by GCP2-3×GFP (those from γ-tubulin complexes) arose in a narrow range of angles centered around 40° to the mother microtubule, with a minority arising in parallel to the mother microtubule (Fig. 3A). These nucleations had a bias to diagonal angles with respect to the cell elongation axis (Fig. 3E), consistent with most mother microtubules lying in transverse orientation. These initiations were initially high in number but fell rapidly after light stimulation (Fig. 3C). By contrast, new growing ends without GCP2-3×GFP (those arising from crossovers, see Fig. 2B) were generated at high angles (most >63°) with respect to one of the partner microtubules (Fig. 3B) and orientated in the same direction as the other partner, consistent with severing that results in a growing new plus end. These initiations peaked ~13 min after light stimulation (Fig. 3D) and were biased strongly to longitudinal orientations relative to cellular coordinates (Fig. 3F). Initiation of growing new ends at crossovers accounted for 83% of all observed longitudinal initiations (71 out of 84 initiations in seven plants) during reorientation (Fig. 3H). Thus, severing by katanin at crossovers appears to be a central engine of cortical array reorientation, quantitatively accounting for the great majority of new longitudinal microtubules.

Fig. 3 Microtubule generation during the reorientation process.

(A and B) Histograms of angles between an existing microtubule and a new microtubule, (A) associated with a GCP2-3×GFP label and (B) associated with crossovers (scored as not associated with GCP2-3×GCP, see text and Fig. 2B). (C and D) Histograms of microtubule initiations over time after blue light stimulation, (C) associated with a GCP2-3×GFP label and (D) associated with crossovers. (E and F) Polar histograms of new microtubule orientations to the cell axis, where the elongation axis is at 90°, (E) associated with a GCP2-3×GFP label and (F) associated with crossovers. (G) Schematic drawing of a dark-grown seedling. The angle reference indicates how the reported angles map onto the cell and plant axis. (H) Bar graph comparing the number of microtubule initiations longitudinal to the cell axis originating from a GCP2-3×GFP label to associated with crossovers (seven plants on 3319 μm2). Of the 212 new microtubules analyzed, 79 were marked by a GCP2-3×GFP–labeled complex and 133 were not. (I) Microtubule-severing–dependent amplification of longitudinal microtubules. The blue overlay shows the paths of all microtubules formed by microtubule severing originating from one single microtubule. Scale bar, 5 μm. See also movie S7.

Generation of New Microtubules by Katanin Has the Properties of an Amplifier

Movie S9

Watch Video

We observed that initiations from crossovers were often not isolated events, but occurred as a series of linked events whereby a single longitudinal microtubule gives rise to a large number of longitudinal microtubules, producing a bifurcating pattern (Fig. 3I and movie S7). When the initial longitudinal microtubule is severed, both the original and the newly generated plus ends typically continue to grow in a longitudinal orientation, creating new crossovers and further opportunities for severing. As the probability of severing is higher for the newer microtubule, this process constitutes a sequential amplifying mechanism to rapidly generate many longitudinal microtubules from a single longitudinal seed.

Katanin Function Is Required for Array Reorientation

To establish whether katanin function is required for microtubule array reorientation, we measured the ability of cortical arrays in the seedling axis to reorient in response to blue light stimulus (see materials and methods). We found that the ktn1-1 mutant (18) was defective in stimulated reorientation (Fig. 4A and movie S8). Transverse order decreased slightly over time (materials and methods and Fig. 4, B and C), but estimates of reorientation speeds were extremely slow (P < 0.001, Mann-Whitney U test; Fig. 4D), and none of the 14 measured cells were able to reorient within the 1-hour observation period, as opposed to 100% of the wild type. These results indicated that katanin function is necessary for rapid cortical array reorientation in response to blue light stimulus.

Fig. 4 Katanin activity is required for microtubule reorientation.

(A) Microtubule reorientation in WT (Ws) and ktn1-1 mutant etiolated hypocotyl cells. Scale bar, 5 μm. See also movie S8. (B) Distributions of microtubule (MT) angles over time (see materials and methods). Time is indicated in (C). (C) Transverse (T) and longitudinal (L) order parameters over time with a quadratic fit (black line) for the cells in (A). (D) Mean reorientation speed (L/min) comparing the wild type (13 cells) and ktn1-1 (14 cells). Error bars denote SEM. The asterisk indicates a significant difference in a Mann-Whitney U test (P < 0.001).

Blue Light Signals Array Reorientation Through Phototropins

Although blue light was employed as a stimulus, it was not known which photoreceptors the light might act through, nor if the activity of those receptors was linked to cortical severing and microtubule initiation at crossovers. We measured array reorientation in mutants for the two major classes of blue light photoreceptors in Arabidopsis, cryptochrome (cry1 cry2) (19) and phototropin (phot1 phot2) (20). We observed a small but significant reduction in array reorientation speed in the cry1 cry2 mutant (P < 0.05, Mann-Whitney U test) and a dramatic reduction in the phot1 phot2 mutant (P < 0.001, Mann-Whitney U test) (Fig. 5, A to C; fig. S5; and movie S10). Thus, we conclude that cortical array reorientation is influenced by cryptochrome signaling but is primarily under the control of the phototropin receptors. The phot2 allele in the double-mutant line is not a null (21), so complete loss of phototropin activity would reasonably be expected to be even more deficient in reorientation response.

Fig. 5 Microtubule reorientation in blue light–receptor mutants.

(A) Distributions of microtubule angles over time in WT and phot1 phot2 mutant etiolated hypocotyl cells (see materials and methods). Time is indicated in (B). (B) Transverse (T) and longitudinal (L) order parameters over time with a quadratic fit (black line) for the cells in (fig. S5). (C) Mean reorientation speed (L/min) comparing the wild type (18 cells), cry1 cry2 (13 cells), and phot1 phot2 (10 cells). Asterisks indicate a significant difference in a Mann-Whitney U test (P < 0.001). Error bars denote SEM. (D) Probability of microtubule crossovers resulting in severing in WT and phot1 phot2 mutant cells differs significantly (P < 0.00001, Fisher’s exact test, two-tailed), indicated by an asterisk (N = 1115 on 656 μm2 in wild type in five plants and N = 616 on 648 μm2 in phot1 phot2 in five plants). Error bars indicate 95% confidence interval (CI).

PHOT1 and PHOT2 Regulate Microtubule Severing

Movie S10

Watch Video

Movie S11

Watch Video

Movie S12

Watch Video

Whereas most new microtubules were initiated from microtubule crossovers in WT cells (57%), the reverse was true in phot1 phot2 mutants, in which only a minority initiated from crossovers (30%), a significant shift (P < 0.00001, Fisher’s exact test; fig. S6D). To address the mechanisms responsible, we asked if the likelihood of initiating a new microtubule at a crossover was up-regulated by phototropin signaling. In phot1 phot2 cells, severing was detected at 36% of crossovers formed during reorientation (220 of 616), whereas in the wild type the frequency was 1.3 times higher at 47% (522 of 1115) (Fig. 5D), a highly significant difference (P < 0.00001, Fisher’s exact test). By contrast, we did not observe a significant difference in the proportion of microtubule-severing events that resulted in a new growing plus end (P > 0.05, Fisher’s exact test, two-tailed), showing no evidence that PHOT1 PHOT2 influences stabilization of the new plus ends. There was also no evidence of a significant difference between WT and mutant cells for the frequency of severing at the preexisting versus new microtubule (63 versus 69% respectively, P > 0.05, Fisher’s exact test). Taken together, these data indicate that blue light signaling through PHOT1 PHOT2 significantly increases the rate of producing new microtubules at crossovers by increasing the likelihood of severing by katanin at these sites. We cannot, however, determine from these data if the increase in severing rate stimulated by blue light is sufficient to trigger reorientation.

We note that because the severing mechanism acts as a sequential amplifier, a modest increase in severing likelihood at each amplification step can have a relatively large effect on the total gain and, thus, the rate of reorientation. This idea is supported by the observation that the number of growing microtubules created at crossovers in the wild type is 2.8 times higher during reorientation than in the phot1 phot2 double mutant, an increase much greater than the 1.3-fold change in severing likelihood per crossover.

The PHOT1 PHOT2 Pathway Influences Microtubule Nucleation

Initiations from sites other than crossovers, 97% of which are associated with detectable γ-tubulin complexes (Fig. 2D), decreased sharply during the course of reorientation (Fig. 3C). In the phot1 phot2 mutant, the initial frequency of these initiations was higher than in the wild type, and this rate decreased less rapidly, yielding a net increase per unit area of 1.5-fold (fig. S6A). We infer from these results that the likelihood of nucleation from γ-tubulin complexes is reduced by a consequence of PHOT1 PHOT2 function, perhaps indirectly, which, together with promotion of katanin severing at crossovers, contributes to the observed switch in prevalence between initiation by nucleation from γ-tubulin complexes or from severing at crossovers in the phot1 phot2 mutant. The proportion of microtubules initiated in parallel to the mother microtubule was significantly higher in the phot1 phot2 mutant (P < 0.01, Fisher’s exact test, two-tailed; fig. S6, B and C). Thus, PHOT1 PHOT2 may also regulate the ratio of branching to parallel nucleation from γ-tubulin complexes, as has been observed for TONNEAU2 (22).

Seeding and Orientating the Amplification of New Microtubules by Severing

The severing mechanism identified here requires a population of longitudinal microtubules to begin with to “seed” the amplification of new microtubules. Branching nucleation starting from transverse microtubules creates mostly diagonal microtubules (Fig. 3E), but a subsequent branching event can create a longitudinal microtubule (fig. S7A and movie S11). The observed PHOT1 PHOT2–dependent preference for branching over parallel nucleation may enhance longitudinal seeding by increasing the likelihood of sequential branching. In addition, diagonal microtubules were also observed to acquire a longitudinal orientation by adopting a curved trajectory of growth (fig. S7B and movie S12). The frequency of creating crossovers and generating new growing ends by severing is highest when the distance between crossovers is shortest, a parameter that is minimized when discordant microtubules grow at right angles to the existing ordered array. Thus, we propose that the existing transverse array acts as an oriented template for the preferential amplification of longitudinally oriented seeds generated by branching nucleation or curved polymer growth (Fig. 6, fig. S7C, and movie S13).

Fig. 6 Schematic representation of katanin-dependent amplification of longitudinal microtubules during cortical array reorientation.

Stability of Reorientation

Movie S13

Watch Video

In principle, reorientation by severing could be reversible. Once sufficient longitudinal order is created, transverse microtubules might be amplified, causing the array to switch back. The substantial drop observed in microtubule nucleation (Fig. 3C) may help suppress such a reversal. Likewise, initiation by severing at crossovers also falls to low levels as reorientation nears completion and the number of crossovers (and, thus, opportunities for severing) drops. The basis for the reduction in nucleation rate remains to be identified. It also remains to be determined if the existing transverse array is actively disassembled after blue light signaling, or if it is simply passively turned over, as generation of transverse microtubules during reorientation is low. Finally, whereas microtubule initiation by severing appears to account for most new longitudinal microtubules, our observations do not exclude contributions to reorientation from other mechanisms.

Phototropism Requires Katanin Function

Blue light signaling through PHOT1 and PHOT2 has been shown to be involved in both phototropism and growth inhibition of the hypocotyl axis (23, 24). Our observation that the ktn1-1 mutant is defective in cortical array reorientation presented us with a genetic test to ask whether microtubule reorientation was required for either response. When we measured growth suppression, we found that the phot1 phot2 mutant failed to show significant reduction in hypocotyl growth speed after the blue light was turned on, but the ktn1-1 mutant showed an immediate reduction in growth speed, with similar kinetics as in WT plants (fig. S8, A and B). These data indicate that katanin-dependent microtubule reorientation is not required for growth inhibition mediated by phototropin signaling, providing genetic confirmation of an earlier correlative study in peas, where growth inhibition was observed to occur significantly more rapidly than cortical microtubule array reorientation (7). Thus, although cellulose synthase trajectories are reoriented in concert with cortical microtubules (3), our results, together with those of Laskowski (7), indicate that such reorientation of the direction of cellulose deposition may not be sufficient to explain rapid growth inhibition.

To determine if phototropism is dependent on katanin activity, we employed a photocurvature assay described by Christie et al. (25) (Fig. 7, A and B). The ktn1-1 mutant showed significantly less 8-hour light-stimulated bending (17.2° ± 1.9°, N = 14 plants) than the wild type (44.8° ± 3.7°, N = 14 plants) (Fig. 7C; P < 0.001, Student’s t test). The growth in length of the hypocotyls of ktn1-1 mutant plants during the hypocotyl-bending assay (0.56 ± 0.03 mm) was not significantly different from that of the wild type (0.47 ± 0.03 mm) (Fig. 7D; P > 0.05, Student’s t test), indicating that the reduced phototropism was not the result of a defect in elongation in the ktn1-1 mutant. The reduction of photocurvature in the ktn1-1 mutant indicates that katanin activity is required to generate normal photocurvature. Microtubule reorientation on the illuminated side of sunflower hypocotyls and maize coleoptiles was observed previously and was thought to contribute to differential expansion and photo curvature (9). Under conditions that stimulate photo curvature in Arabidopsis, we also found that cortical arrays reoriented on the illuminated side of the axis in a PHOT1 PHOT2–dependent fashion (fig. S8C). Given these observations and the demonstrated linkage between microtubule orientation and regulation of directional cell expansion under many conditions (5), our genetic results are consistent with the possibility that phototropin-stimulated microtubule reorientation on the illuminated side of the axis is an important contributor to phototropism of the hypocotyl axis.

Fig. 7 Microtubule reorientation and the phototropic response.

(A) Seedlings of the wild type (Ws) and ktn1-1 before and after photocurvature assay. BL, blue light. Scale bar, 1 mm. (B) Pictogram of the measurement of photocurvature (θ). (C) Mean angle of curvature in the wild type and ktn1-1. Error bars indicate SEM (N = 14 for both wild type and ktn1-1). The asterisk represents a significant difference between the wild type and ktn1-1 (P < 0.001, Student’s t test, two-tailed). (D) Mean hypocotyl growth of the wild type and the ktn1-1 mutant during the photocurvature assay [growth as calculated by subtracting the measured hypocotyl length between yellow carets from the measured hypocotyl length between red carets in (A)]. Error bars indicate SEM. N = 50 for the wild type and 43 for ktn1-1.


The mechanism revealed here for microtubule array reorganization stands in interesting contrast to current models. Live-cell imaging and modeling studies have indicated that cortical microtubule arrays in higher plants have self-ordering properties (2632). Differential regulation of microtubule stability tied to location in the cell has been proposed to act on self-ordering of the array as a whole to produce specific array patterns, an idea supported by studies of microtubule stabilization at specific cell edges by CLASP (33) and destabilization in defined plasma membrane domains by MIDD1 (34). By contrast, katanin-dependent amplification does not rely on localized regulation of stability, but depends on the preexisting microtubule organization. The new order created does not depend on global self-ordering but arises from the accumulation of independent amplification events. Remodeling of cortical arrays in response to application of plant hormones has also been described recently in hypocotyl cells (35, 36). The mechanism has not been identified, but the progressive nature of new ordering (36) indicates that the pathway is distinct from that caused by blue light signaling.

Cortical arrays in plant cells feature distributed γ-tubulin nucleation complexes that are recruited in a distributed fashion, primarily to existing microtubules (10, 17). Our observations in this study reveal a second means of generating new microtubules in these acentrosomal arrays, based on microtubule severing by katanin. Like nucleation from γ-tubulin complexes, the location and geometry of microtubule generation by this alternative mechanism is tied to the distribution of existing microtubules. This mechanism was not only important in its prevalence but was also a target of signaling and actually dominated during our observations of cortical array reorientation. It will be interesting to assess the contribution of this mode of microtubule generation in the maintenance and transitions of other arrays in higher plants.

Katanin function is important for diverse cellular and developmental processes across a broad range of biological diversity, including neuronal development, spindle assembly and function, mitosis, cell migration, turnover of cilia and flagella, cell morphogenesis, and cell wall biogenesis (3741). However, the specific role or roles of microtubule severing in supporting those processes are typically not well understood, often due to challenges in performing dynamic studies because of the high density of the arrays under study (42). In some cases, severing appears to play a relatively straightforward role in separating microtubules from their nucleation centers (10, 43, 44), facilitating microtubule turnover. Observations in the meiotic spindles of Caenorhabditis elegans have led to the proposal that katanin-mediated severing may also be used to build new organization in acentrosomal arrays by making new microtubules (42, 45, 46), a possibility that has also been suggested in neuronal arrays (42). The low-density two-dimensional geometry and optical accessibility of plant cortical arrays have allowed us to test this proposal by visualizing katanin recruitment, individual severing events, and their outcomes. These experiments have revealed a previously undiscovered and unexpected mechanism for selectively amplifying a subpopulation of microtubules based on severing at microtubule intersections. Our studies further highlight plant cortical arrays as useful systems for investigating the roles of widely conserved functions, such as severing and nucleation, in the dynamic reorganization of acentrosomal microtubule arrays.

Materials and Methods

Plant Material

All live-cell imaging experiments were performed in 3-day-old dark-grown etiolated hypocotyls of A. thaliana. Three constructs were used for imaging of cortical microtubules: 35S:YFP-TUA5 (YFP, yellow fluorescent protein) (25), 35S:GFP-TUB6 (47), and 35S:mCherry-TUA5 (4). The wild type and the cry1-104 cry2-1 (19) and phot1-5 phot2-1 double-mutant lines (48) expressing the YFP-TUA5 marker were in the Columbia (Col0) ecotype. The wild type and the ktn1-1 mutant expressing the GFP-TUB6 marker were in the Wassilewskija Ws ecotype (47). WT Arabidopsis plants expressing YFP-TUA5 and mCherry-TUA5 were in the Col-0 ecotype. γ-tubulin complexes were visualized by expression of GCP2-3×GFP from native upstream sequences (10).

The GFP-KTN1 construct was generated from the katanin p60 subunit (At1g80350) genomic region including a 1005–base pair region 5′-upstream from the initiation ATG codon and a 1039–base pair region 3′-downstream from the stop codon. A sequence encoding smRS-GFP (49) inserted at SmaI and NaeI sites, which were introduced before the initiation ATG codon of KAT1 by polymerase chain reaction. The insert was excised by NotI and was introduced into the binary vector pBIN19. The resulting binary vector was used to transform ktn1-2 mutant plants (Col0 ecotype) (10, 15). mCherry-TUA5 expressing plants were then crossed to the GFP-KTN1/ktn1-2 lines, and progeny expressing both reporters were analyzed.

Plant Growth

Seeds were surface-sterilized, stratified for 3 days at 4°C, and sown on 1% agar containing Hoagland’s no. 2 basal salts at pH 5.7 for all experiments except the photocurvature assays, for which we used 0.5× Murashige and Skoog media (50) with 1% (w/v) sucrose and 1% (w/v) agar at pH 5.6. After 1 hour of light exposure on a bench top, plates were incubated in a near-vertical position at 22°C and wrapped in foil for 60 to 72 hours.

Specimen Mounting for Confocal Microscopy

Seedlings were mounted under red safelight conditions (Paterson PTP760U Safelight) to prevent de-etiolation. Seedlings were gently placed on a cover slip in sterile water and affixed under a 1-mm-thick 1% agarose pad.


In experiments where microtubule reorientation and/or GCP2-3×GFP were imaged, we used a spinning-disk confocal microscope equipped with a CSU-X1 spinning-disk head (Yokogawa) on a Nikon Eclipse Ti body with Perfect Focus system (Nikon) (4, 51). In experiments where mCherry-TUA5 and GFP-KTN1 were imaged, we used a CSU-X1 spinning-disk confocal head mounted on a Leica DMI6000B microscope with Adaptive Focus Control (Leica) and a 100× Plan Apo 1.4 N.A. oil immersion objective. GFP was excited at 488 nm and mCherry at 561 nm using a multichannel dichroic filter (Di01-T405/488/568/647, Semrock) and an FF01525/50 or an FF01605/64 bandpass emission filter (Semrock) for GFP and mCherry, respectively. Images were acquired with an Evolve electron-multiplying (EM) charge-coupled device camera (Photometrics) at an EM gain of 300, controlled by Slidebook software (Intelligent Imaging Innovations).

Confocal Imaging

To ensure that the imaging and light conditions were the same for every experiment, we measured the laser power of the 491-nm laser (and 561 nm for the mCherry excitation) at the optical-fiber output that goes in the spinning-disk head and set it to 8.2 mW, which corresponded to 1.05 mW at the back focal plane of the objective. Images used for measuring the reorientation speeds of cells were acquired with a 300-ms exposure time every 10 s over 60 min and every 5 s over 30 min for the movies used for analyzing microtubule initiations and severing. The mCherry-TUA5 GCP2-3×GFP plants were imaged with every 5-s interval over 30 min with a power of ~1.05 mW at the back focal plane of the objective and 500-ms exposure time for both the 491- and 561-nm lasers at the optical fiber. In experiments where mCherry-TUA5 and GFP-KTN1 were imaged, we used 800-ms exposure every 5 s over 30 min with ~5 mW at the end of optical fiber. For measurements of initiations and severing, rectangular sample regions were defined in each image series, and detectible events of interest (initiations, crossover formation, and shortening of new plus ends) were exhaustively assayed throughout the duration of the time series.

Image Analysis for Microtubule Orientations

To quantify microtubule orientation and order, we developed an automated image-analysis tool in the form of an ImageJ plugin, LOCO (local orientation). LOCO identifies the orientation of high-contrast linear objects in images, such as fluorescently labeled microtubule arrays, assigning each pixel above a threshold level with a value for the local orientation of the signal, as defined by a configurable rotating kernel. The plugin allows the user to define the kernel dimension, the angular interval, and the threshold value. In our analysis, we chose a width of 3 pixels, a length of 21 pixels, an interval of 9° (corresponding to 20 angles over a 180° rotation interval), and a threshold value determined by the default ImageJ threshold (52). The chosen pixel width of 3 was comparable to the width of the typical microtubule signal and the length of 21 made the orientation filter sensitive to orientation of the signal with no observable border effects. Images were padded with a border of zero-value pixels at a width equivalent to (k – 1)/2, where k is the kernel length. At each pixel of the image, the kernel was rotated every 9° over the 180° range, and pixel intensities were summed over the kernel. Intensities were calculated using bilinear interpolation. The angle that resulted in the highest integrated kernel intensity was returned as the preferential angle value for that pixel. As visual output, the plugin produces an image of pixel angle values above the threshold level (movie S9). These data sets were processed further in MATLAB (MathWorks) to create heat maps of angular distributions over time and calculate order parameters. The LOCO plugin is available at

Microtubule Reorientation Analysis

The image analysis produced the calculated local orientation about all pixels above the threshold value over 20 bins. To determine the moment in time when the longitudinal microtubule order becomes dominant over the transverse order, we defined two filter functions: T for transverse order and L for longitudinal order (51). With 90° defining the transverse axis, these values were calculated as follows: For T, the fraction of angular values in a transverse bin 45° wide centered at 90° was given a weighting of 1, the fraction at all other orientations given a weighting of –1/3, and these products summed. The weightings were chosen so that an isotropic distribution of angles yields a value T = 0, and a distribution contained entirely within the transverse bin yields a value T = 1. Likewise, L was constructed using a longitudinal bin 45° wide centered at 0°. We fitted the values of T and L over time using a quadratic polynomial. We wanted to focus on the buildup of the longitudinal order, so we used the average increase in longitudinal order over time as the measure for the speed of reorientation. Because L levels off as longitudinal order is obtained, we calculated this value from the start of acquisition until L reached 0, an interval over which L increased approximately linearly. In cases where L did not reach zero, we divided the total increase in L by the total observation time. So in short, we defined the reorientation speed as LT = 0/TL = 0. Movies that did not show clear transverse order at the start (T0 < 0.1, ~10% of the plants) were excluded, as they did not provide information on reorientation dynamics. These calculations and procedures were incorporated into a MATLAB script to enable rapid processing of large data sets and limit observer bias.

Microtubule Initiation Analysis

As a visual aid for detecting new microtubules, we performed a time-phased image subtraction with a phase shift of three frames on the mCherry-TUA5 signal and selected the positive and negative difference using MATLAB. The positive difference corresponds to polymer gain, and the negative difference corresponds to polymer loss. This manipulation results in comets reminiscent of fluorescent labeling of plus-end binding proteins (+TIPs). We made the comets from the positive-difference image green and added them to the grayscale microtubule images. We found that a difference of three frames (or 15 s) was optimal for the visibility of the growing microtubule ends. The negative-difference images were processed in the same way as red comets, as shown by the overlay in movie S1. Like growing ends labeled with +TIPs, the comets produced by time-phased subtraction are easily discerned and quantified, but this method has a couple of distinct advantages over +TIP labeling. First, when using +TIP labeling, it can be challenging to distinguish new microtubule initiations from rescue events. Second, the organization of the microtubule array needs to be inferred by tracking the trajectories of the labeled plus ends. However, because losses at both plus and minus ends are not seen, the actual distribution of microtubule lattices cannot be known at any given time by plus-end tracking alone. The time-phased subtraction technique addresses both of these challenges because the lattice signal and loss events are also imaged. Further, the time-phased subtraction technique avoids unwanted perturbation of microtubule dynamics and behaviors caused by overexpression of a +TIP protein.

For all new microtubules, we scored the position, time, whether it originated from a preexisting microtubule, the angle of the preexisting microtubule, the angle of the new microtubule relative to the existing microtubule, the angle of the new microtubule relative to the cell axis (as indicated in Fig. 1B), whether the new microtubule originated from a microtubule crossover, and whether GCP2-3×GFP was present in the GFP channel (in the double labeled line). When a new microtubule initiated from a crossover, the orientation was always within a narrow range of one of the microtubules before the severing event occurred. We therefore calculated the angle between the newly created plus end and the microtubule, with the microtubule with the highest angle to that.

For the crossover analysis, we marked the position and moment of crossover creation, the moment of crossover disappearance, and whether a GFP-KTN1 signal was present (in the double labeled lines). When a crossover disappeared, we asked whether severing of the new microtubule had taken place and whether that microtubule started off in a growing or shrinking state. We asked the same questions for the preexisting microtubule at the crossover.

Hypocotyl Elongation and Photo Curvature

To measure hypocotyl elongation, 3-day-old dark-grown seedlings germinated on vertically oriented plates were imaged with a red safelight for illumination (Paterson PTP760U Safelight) and imaged with a Nikon D80 camera with a macro lens at 1-s exposure time, with a 5-min time interval. After 1 hour, we turned on a blue light (10 μmol m−2 s−1) with a diode array (EagleLight) parallel to the plate. We made kymographs along the elongation axis of the hypocotyls to measure the growth speeds before and after blue light induction.

Photo curvature was analyzed according to methods described in (25). Plant seedlings were germinated in the light for 3 days and illuminated from the side by monochromatic 2–μmol m−2 s−1 blue light (EagleLight) for 8 hours after 2 days dark acclimation.

Supplementary Materials

References and Notes

  1. Acknowledgments: D.W.E., J.J.L., M.N., and V.K. designed experimental strategy. J.J.L., M.N., A.H., and V.K. carried out experiments. J.J.L., M.N., A.H., and D.W.E. analyzed data. K.S., R.G., J.J.L., B.M.M., and D.W.E. designed and created image-analysis tools. D.W.E., J.J.L., and M.N. wrote the manuscript. B.M.M., A.H., R.G., V.K., T.K., and A.M.C.E. edited the manuscript. We thank W. Briggs, A. Murphy, R. Hangarter, E. Spalding, T. Hashimoto, M. Janson, and S. Shaw for advice and discussions. This work was supported by NSF award 1158372 (D.W.E.); the Carnegie Institution for Science (D.W.E.); the research program of the Stichting voor Fundamenteel Onderzoek der Materie (FOM), which is financially supported by the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO) (B.M.M.); the European Union NEST program CASPIC award 28974 (J.J.L. and K.S.); TOYOBO BIOFOUNDATION (M.N.); and the Human Frontier Science Program (M.N.).
View Abstract

Navigate This Article