CLASP Modulates Microtubule-Cortex Interaction during Self-Organization of Acentrosomal Microtubules

Plant Materials and Growth Conditions

Seeds were cold-treated after planting for 48 h to synchronize germination. Seedlings were grown aseptically on Hoagland's medium solidified with 1.2% agar at 23°C under continuous light. Cotyledons were excised from 7 to 12 d seedlings and placed in water between a slide and coverslip to enable imaging of the adaxial (top) surface at 24°C. The phenotypes described here were observable at all stages of development (4–12 d). No more than two cells were imaged per cotyledon to minimize phototoxicity. Wild-type seeds expressing GFP-β-tubulin 6 from Arabidopsis (Nakamura et al., 2004) were obtained as a gift from Dr T. Hashimoto (Nara Institute of Science and Technology, Japan) and crossed to clasp-1 plants (SALK accession no. salk_120061).

Tissue Preparation and Microscopy

Images were acquired on a Zeiss Axiovert (Zeiss, Göttingen, Germany) microscope equipped with an axiocam HRmCCD camera, driven by Axiovision software (Zeiss). Bandpass filters for GFP were 460–480-nm excitation and 505–530-nm emission. Typical light exposure regimens were 1–1.5-s exposures at 30% arc lamp intensity (Zeiss FluoArc), at 5-s intervals. Photobleaching was negligible under these conditions. Individual cells were observed for no more than 5 min, because MT detachment activity was sensitive to photodamage. Confocal imaging was performed with a 40× plan-apochromatic water immersion objective mounted on a Zeiss Pascal, using the 488-nm line from an argon laser. Typical scan times were 4 s, using a line averaging of two. Slice thickness was 1.5 μm. For latrunculin B treatments, a 20 μM solution in 1% DMSO was prepared from a 2 mM stock in 10% DMSO. Cessation of streaming was typically observed at 5–10 min after application. DMSO, 1%, alone had no obvious effect on detachment activity or MT organization.

Image Analysis and Statistical Analysis

Image analysis was performed using ImageJ software (http://rsb.info.nih.gov/ij/). Figures were assembled in Corel Draw (www.Corel.com; Corel Systems, Ottawa, ON, Canada). Statistical analysis was performed using Microsoft Excel (Microsoft, Redmond, WA).

Loss of CLASP Generates Hyper-Parallel Cortical MT Arrays That Exhibit Reduced Cortex Anchoring

To understand the role of CLASP in the spatial organization of cortical MT arrays, we compared MT behavior in wild-type and clasp-1 mutant Arabidopsis thaliana plants, which contain a T-DNA insertion that knocks out expression of the single-copy CLASP gene (Ambrose et al., 2007; Kirik et al., 2007). In agreement with our previous immunofluorescence data reported for roots (Ambrose et al., 2007), as well as for GFP-tubulin in hypocotyl cells (Kirik et al., 2007), MTs in most clasp-1 cells were well organized into parallel arrays. Analysis of MT orientation in living cells expressing GFP-tubulin (GFP-TUB6) revealed that in cotyledon and leaf epidermal cells, clasp-1 MTs exhibited a greater degree of parallel order than those of wild type (Figure 1). In contrast to the varied MT orientations present within individual wild-type cells, clasp-1 cells typically possessed one predominant MT orientation throughout the cell (Figure 1, A and B). Quantification of MT angles in individual cells showed a narrowing of the angular distribution in clasp-1 compared with wild type (Figure 1B). Consistent with the increased MT parallelism, a higher proportion of MT plus ends were found to be growing along preexisting MT tracks in clasp-1. From time-lapse image series obtained with a high-resolution CCD camera, we determined that clasp-1 mutants have an approximate threefold increase in the ratio of MT plus ends growing along other MT tracks compared with those growing freely (Figure 1C).

In clasp-1 cells, MTs displayed a conspicuous behavior, wherein MTs partially detached from the cortex over large lengths and durations, deviating from their normally linear growth trajectories. The detached portions underwent extensive waving, bowing, and changes in orientation, particularly when exposed to the forces of cytoplasmic streaming (Figure 2).

These behaviors can be most easily appreciated in the accompanying time-lapse movies (see Supplementary Movies S1–S9). Figure 2A shows a typical reorientation, in which the leading end of a growing MT dissociates and undergoes a large change in orientation before reassociation with the cell cortex (see Supplementary Movie S4). MTs were frequently observed to partially reattach to the cortex in one or more locations. Figure 2B shows an example of this, in which the leading end of the detached region reassociates with the cortex and begins growth in a linear trajectory, whereas the middle segment of the MT remains free (Figure 2B; see Supplementary Movie S5). Extensive MT distortions between the two points of cortical anchoring are apparent (arrowheads). Partial reattachment points were often transient, or upon partial reattachment, the initial region of attachment was propagated along the MT, resulting in larger regions of attachment.

Cortex detachments were occasionally observed in wild-type cells, albeit much less frequently, and involving less extensive stretches of MTs than those observed in clasp-1 cells (Figure 2, C–E). Specifically, MT detachments in clasp-1 were 3.2 times more frequent, were 1.6 times longer in duration, and the detached lengths were twice as long compared with those observed in wild-type cells (Figure 2, C–E; Table 1). The observed cortical detachments in wild type suggest that the clasp-1 phenotype is an exacerbation of a normal behavior.

The enhanced cortical MT dissociation in clasp-1 was observed in all cell types examined (hypocotyl epidermal, lateral root cap, and cotyledon epidermal) and was most easily documented in cotyledon epidermal cells, which are shown here. Although most MTs in clasp-1 cotyledon epidermal cells resided within parallel arrays, most of the detachment behavior was observed in the subpopulation of MTs with discordant orientations relative to the predominant array. On the basis of the enhanced parallelism and the observation that MTs frequently realigned in the direction of the predominant cytoplasmic flow upon reattachment, we initially hypothesized that the enhanced parallelism observed in clasp-1 mutants is due to cytoplasmic streaming-dependent flow alignment. The hypothesis, however, is not supported by the following data. First, in many cells, the prevailing direction of cytoplasmic streaming did not match the predominant MT orientation, often flowing at oblique or right angles to the MT alignment axis, or consisting of multiple flow vectors in different areas of the cortex. Second, MT arrays remained parallel upon elimination of acto-myosin–based cytoplasmic streaming with the drug latrunculin B, even after prolonged treatments of >24 h (Supplementary Figure S1). MTs in latrunculin B–treated clasp-1 cells still detached from the cortex, although they did not show the characteristic lateral displacements observed in untreated cells (Supplementary Figure S1). Furthermore, streaming velocities did not appear to be affected in untreated clasp-1 cells. These data show that streaming is not required for the maintenance of MT parallelism or that it plays only a minor role.

Partial Cortical MT Detachment Enables Interactions between MTs and Bundle Formation

Partially detached MTs in clasp-1 often waved in the cytoplasmic stream until encountering another partially detached or a cortex-associated MT, resulting in the two MTs associating laterally to form a bundle (Figure 3, A and B; see also Figure 2A).

This search-and-capture–based MT bundle formation occurred in two distinct ways: 1) polymerization-dependent bundling, in which the detached plus-end encounters and then grows along another MT track (Figure 3A; see Supplementary Movie S6); and 2) polymerization-independent bundling, which can occur either by zippering, in which the detached segment (i.e., plus ends and/or interior portions) encounters and then progressively anneals behind the initial contact point (brackets in Figure 3, A and B, and brackets in Figure 2A), or by abrupt uniform lateral annealing of the two MTs (Figure 3C, brackets; see Supplementary Movie S8). The zippering mechanism is similar to the partial cortical reattachment described above, where cortex attachment is propagated from an initial point. Two unattached MTs frequently coaligned into a bundle upon encounter (Figure 3B; see Supplementary Movie S7). When two unattached MTs approached one another at different angles and interacted, the region of MT overlap typically straightened as the individual MTs grew together (Figure 3B, brackets). We also observed frequent detachment of MT bundles (as determined by filament width and fluorescence intensity) in clasp-1 (Figure 3B). The example in Figure 3C shows a detached bundle “snapping” together with another MT.

These conspicuous MT interactions in clasp-1 prompted us to quantify MT bundle formation, which we define here as the annealing of two or more MTs into a single linear filament. To avoid erroneous inclusion of instances where MTs coincidentally grew in overlapping trajectories, bundling was only considered to have occurred when alignment was accompanied by a clear change in growth trajectory or position of one or both of the aligning MTs.

Our analysis identified a strong correlation between the frequency of MT detachment and MT bundle formation. Bundle formation events were roughly three times more frequent in clasp-1 than in wild type (Figure 4A), which correlates closely with the 3.2-fold increase in detachment frequency (Figure 2C). Of these bundling events, 85% in clasp-1 were associated with a detachment of one or both of the aligning MTs, compared with 52.5% for wild type (Figure 4B). In the absence of CLASP, detached portions of MTs were also far more likely to reattach to the cortex via MT bundling rather than to reassociate freely with the cortex. In clasp-1, over three quarters of cortex reattachment events resulted in bundling upon reattachment, compared with roughly half in wild type (Figure 4C). Increased bundle formation in clasp-1 is further supported by the observation that freely growing MTs, regardless of their cortical attachment status, spent less time before coaligning with other MTs in clasp-1 compared with wild type (73.1 ± 4.1 s vs. 85.8 ± 4.4 s, p = 0.018, t test).

What is the basis for the increased prevalence of MT bundle formation in clasp-1 cells? One possibility is that in the absence of CLASP, MTs bind each other more readily upon interaction, perhaps through increased availability of sites for cross-linking proteins. A second possibility is that the reduced cortical attachment in clasp-1 enables MTs to explore a larger space, thereby increasing the incidence of MT encounters that are favorable for bundle formation. To differentiate between these two explanations, we quantified detachment events and bundling in wild type, where there is a natural range of detachment activity from cell to cell.

First, we drew a correlation between degree of detachment and bundle formation on a cell-to-cell basis. Wild-type cells that showed the highest degree of detachment activity also exhibited the highest frequency of bundling. The proportion of bundle formation events associated with MT detachment rose to as high as 80–90% in cells with the greatest degrees of detachment activity, compared with the above-noted wild-type average of 52.5%. Conversely, this proportion dropped to 10–20% in cells with the lowest detachment activity. Second, we were also able to draw a positive correlation between the degree of MT detachment and coalignment activity on an individual MT basis in wild-type cells. As shown in Figure 4D, longer MT detachment lengths are more likely to bundle with other MTs upon reattachment. Specifically, MTs that detached and bundled with other MTs were detached over a 31% greater MT length compared with those that did not bundle with other MTs upon cortex reassociation (Figure 4D).

Taken together, our data suggest an indirect role for CLASP in modulating MT–MT interactions, whereby maintaining MTs in close association along their entire lengths with the cell cortex reduces their exploratory capacity. When CLASP's function is reduced, larger partial MT detachment lengths allow exploration of larger swaths of cortical cytoplasm, thereby increasing the probability of finding and freely coaligning with other MTs.

CLASP Restricts the Angle of Incidence of MT Coalignment

On the basis of the findings that interactions between MTs may contribute to the self organization of noncentrosomal cortical MTs in plants and animals (Dixit and Cyr, 2004a; Reilein et al., 2005) and the observations here of increased bundle formation in clasp-1, we next sought to determine whether the outcomes of MT–MT interactions are affected in clasp-1 cells. As documented by others (Shaw et al., 2003; Dixit and Cyr, 2004b), we also observed that growing MTs encountering an obstacle MT at small angles (<40°) frequently coalign to form a bundle. To distinguish this coalignment behavior from other types of bundling events, we define coalignment here specifically as the interaction and coalignment of a growing MT plus end with the obstacle MT it encounters. In wild-type cells this was the predominant mechanism of bundle formation, whereas in clasp-1, it was accompanied by the numerous detachment-associated mechanisms of bundle formation described above. Quantifying MT coalignments at first proved difficult in the steady-state arrays of clasp-1 because many MTs were parallel to one another, thereby making collisions between freely growing plus ends and other MTs relatively infrequent. To achieve a situation wherein clasp-1 was comparable with wild type, we rapidly depolymerized MTs with cold (−20°C, 4 min) and then observed their repolymerization upon return to room temperature (24°C). Immediately after cold treatment, MTs in clasp-1 and wild type were completely depolymerized or retained a few short remnants of MT bundles (where bundles resided before chilling; Figure 5A; see Supplemental Movie S9). On return to room temperature, cells exhibited normal, rapid cytoplasmic streaming and MT behavior and returned to full polymer status within <10 min, indicating the treatments were not detrimental to the cells (Figure 5A; see Supplementary Movie S9). At larger encounter angles, clasp-1 MTs were more capable of coaligning than wild-type MTs (Figure 5B). MT coalignments in clasp-1 occurred at up to 105° initial encounter angles, with a significantly higher mean coalignment angle of 35.2° for clasp-1 and 23.8° for wild type (Figure 5B). Notably, large-angle MT coalignment events were also typically accompanied by partial detachment of the coaligning MT, consistent with our observations in non–cold-treated cells that MT detachment facilitates more promiscuous bundling activity.

Cortical MT Detachment Facilitates MT Coalignment

Previous reports have described MT coalignment between cortex-associated MTs (Shaw et al., 2003; Dixit and Cyr, 2004b). We observed that in both wild type and clasp-1, MT coalignments are often associated with varying degrees of cortical detachment. In both wild type and clasp-1, encounter of a growing plus end with an obstacle MT is frequently followed by a brief dissociation from the cortex of the leading end of the growing MT. Importantly, for shallow angle encounters, this obstacle-associated detachment often results in coalignment, as the detached portion rapidly pivots into alignment with the obstacle MT (Figure 6A). Similarly, upon obstacle-based detachment, the MT may simply reassociate with the cortex and continue growth in the original trajectory or undergo reorientation during detachment and reassociate in a new growth path (Figure 6B). This was more typical of steep angle collisions, which have previously been shown to result in crossover of the obstacle MT (Dixit and Cyr, 2004b; Wightman and Turner, 2007).

In clasp-1, MT–MT encounters produced many of the large and prolonged detachment episodes described above, frequently resulting in coalignments with nearby nonobstacle MTs (Figure 3A). These data demonstrate that encounter of a growing MT plus end with an obstacle MT can lead to partial cortical detachment of the leading end, which subsequently allows increased angular freedom of the searching MT.

Function of CLASP Sidewall Binding versus Plus-End Tracking

CLASP's role in cortex attachment may be conserved across eukaryotic taxa. Centrosome-dependent MT organization is typical in proliferating cells but many differentiated animal cell types have extensive noncentrosomal MTs as well as MTs that associate along their length with the cortex (Keating and Borisy, 1999; Bartolini and Gundersen, 2006). Interactions between MTs are important in the MT arrays found in association with the basal cortex of epithelial cells (Reilein and Nelson, 2005; Reilein et al., 2005). Interestingly, in these cells the plus-end tracker EB1 also associates along the length of MTs, in contrast to its characteristic plus-end binding (Reilein and Nelson, 2005). In this manner, the regulation of MT plus-end versus sidewall binding of MAPs such as EB1 and CLASP represents a way to modulate cortical association of MTs (i.e., end-on attachment vs. lengthwise attachment).

The binding of CLASP to the full length of MTs in plants is informative as to the general role of CLASPs in eukaryotic cell MT-cortex association. The observation that the lattice binding of CLASP is differentially regulated in animal epithelial cells suggests that it may be an important determinant in regulating cortical association of MTs. In a simplistic model, sidewall labeling of CLASP (and other proteins such as EB1) results in lateral cortex association (as in higher plants and possibly in migrating cell leading-edge lamellae), whereas restricting CLASP to the plus ends favors the end-on MT-cortex association found in many animal cell types. Although we and others observed a slight enrichment of CLASP at MT plus ends (Ambrose et al., 2007; Kirik et al., 2007), it remains to be seen whether this localization can be modulated in plants as well, perhaps in different cell types or stages of the cell cycle. One situation where one may predict a stronger plus-end association of CLASP in higher plants is after cell division, when MTs are nucleated on the nuclear envelope and grow radially through the cytoplasm toward the cortex.

In animal cells, CLASPs interact with two membrane bound proteins, LL5β and ELKS (Lansbergen et al., 2006). BLAST searches indicate an absence of clear orthologues for these in the Arabidopsis genome. Identifying additional molecular components involved in plant cortical MT association will be of interest in the future.