A. thaliana wild type Columbia (Col-0) ecotype and ktn1-2, a T-DNA mutant were used. For germination, Col-0 and mutant seeds were surface sterilized, plated on 0.8% w/v Phytagel® solidified 1/2 Murashige and Skoog medium (1/2 MS; Duchefa) with 1% w/v sucrose, stratified for 1–4 days at 4°C and subsequently transferred to environmental chamber with controlled light/dark cycle, temperature, and humidity.

Unless stated otherwise, all common chemicals were from Sigma and were of analytical grade. FM4-64 was from Invitrogen and it was used at a final concentration of 5μg/ml diluted in half-strength liquid MS medium.

For live cell imaging Col-0 and ktn1-2 mutants stably expressing a GFP-TUA6 marker were used. For generating transgenic ktn1-2 line with GFP-TUA6, ktn1-2 homozygotes (Nakamura et al., 2010) were crossed with Col-0 plants stably transformed with a 35S::TUA6:GFP construct (Shaw et al., 2003). For imaging purposes, 7–10 day old seedlings grown from F2 seeds were used after selection for obvious ktn1-2 phenotype and expression of GFP.

For live imaging of microtubules in the ktn1-2 mutant we used four different Zeiss microscopy platforms (Zeiss Microscopy, Oberkochen, Germany) including an LSM710 spectral CLSM, a Cell Observer, spinning disc, an LSM880 with Airyscan and an Elyra PS.1 unit for SIM (Komis et al., 2014, 2015). For documentation of cortical microtubule dynamics we used either SIM coupled to a PCO. Edge 5.5 sCMOS camera (Komis et al., 2015) using the 488 nm line of an Argon laser for excitation and appropriate filter cube for emission, or the LSM710 with 4-line averaging and pinhole set to 1 Airy unit for GFP coupled to a 32 GaAsP detector. For documenting mitotic spindle dynamics we rather used a Cell Observer spinning disc system with a 25 mW multiline Argon laser (which was set to 50% of laser power of the 488 nm line for every experiment), coupled to an Evolve 512 EM CCD camera, setting laser power and camera exposure time (typically 100 ms) to the minimum possible to allow acquisition of 3D time series. Alternatively, we have used also LSM880 with Airyscan and single photon excitation with the 488 nm line of a multiline argon laser with the same specifications as mentioned before (25 mW for the 488 nm line) and a 32 GaAsP detector. Due to the superior light collecting capacity of the Airyscan and the sensitivity of the GaAsP detector, laser power never exceeded 1% of the laser. In this case samples were either scanned with the Fast mode to allow sufficient temporal resolution during 3D-time acquisition, or with the super-resolution mode to allow optimum 3D resolution for acquiring single Z-stacks. To avoid phototoxicity perturbations and temporal delays during the documentation of the mitotic progress, we preferably used the Airyscan CLSM and the spinning disc microscope for recording mitotic cells, since they allowed the most efficient light collection compared to other platforms and also a fast scanning mode of the sample. The ability of cells to progress through mitosis and complete cytokinesis without disturbances was considered as proof of lack of phototoxicity for the duration of each respective experiment.

Images acquired with the above microscope systems were post-acquisition analyzed. Angular distribution of cortical and PPB microtubules was deciphered by means of Cytospectre (Kartasalo et al., 2015). Using this software, it is possible to decipher the angular distribution of cortical microtubules irrespectively of their polarity (since GFP-TUA6 labels the microtubule lengthwise) and thus deduce their net orientation as “transverse” (i.e., perpendicular to the cell axis) at angles of 90° or 270°, “longitudinal” (i.e., parallel to the cell axis) at angles of 0° or 180° and random at angles 0°<φ <180°. The software also provides an estimate of the ratios of polymers per angle (0–1 at 0.2 increments; concentric cycles in the respective graphs).

For quantifying fluorescence skewness (a measure of microtubule bundling) we used previously published macros (Higaki et al., 2010) which were developed for Image J (http://rsb.info.nih.gov/ij/) on SIM images of Col-0 or ktn1-2 petiole or cotyledon epidermal cells carrying a GFP-TUA6 molecular marker for microtubules. These macros belong to the formerly known KBI suite from Hasezawa lab and can be now found in LPixel ImageJ Plugins (https://lpixel.net/products/lpixel-imagej-plugins/).

For microtubule dynamic analyses, kymographs were generated using the appropriate Kymograph plugin of Zeiss Zen Blue 2014 (Zeiss Microscopy, Oberkochen, Germany), or the MultipleKymograph plugin for Image J (https://www.embl.de/eamnet/html/kymograph.html). From such kymographs, variables such as end-wise growth and shrinkage rates, catastrophe, and rescue frequencies were calculated. In particular, growth and shrinkage rate were calculated from the growth and shrinkage slope of the respective kymograph. Catastrophe or rescue frequencies were calculated by dividing the total number of events recorded (catastrophes and rescues, respectively) by the total time spent in the growing (for catastrophes) or shrinking (for rescues) phase during which these events were counted (Verde et al., 1992) for each microtubule recorded. Severing events in Col-0 and ktn1-2 mutants were measured from SIM videos and were expressed as events × 10⁻³/μm²/min.

Wherever applicable wild type and ktn1-2 microtubule dynamics parameters were compared by two-tailed Student's t-test. Statistical significance was deemed when p < 0.05.

Petiole epidermal cells of Col-0 are elongated and as such they favor the predominant transverse orientation of cortical microtubules (Figures 1a,b). KATANIN 1 depletion in the ktn1-2 mutant, promoted the isotropic growth of petiole epidermal cells and prevented the transverse orientation of cortical microtubules which rather occupied the cell cortex at random orientations (Figures 1c,d). Further inspection of cotyledon epidermal cells, showed that by comparison to Col-0 where microtubules where randomly oriented (Figure 1e), ktn1-2 mutants exhibited similar patterns of organization. The only difference in this case was the observation of multiple branched nucleation events (Figure 1f, inset) compared to Col-0 where branching was more sparse (Figure 1e, inset).

KATANIN 1 depletion also affected the degree of microtubule bundling in petiole and cotyledon epidermal cells. The above measure was quantified in terms of skewness (Figure 2). In the first case, it was found that bundling of cortical microtubules was significantly reduced in both petiole and cotyledon epidermal cells of ktn1-2 (Figures 2c–f) as compared to Col-0 (Figures 2a,b,e,f).

Documentation of microtubule plus and minus end dynamics showed that by comparison to Col-0 (7.41 ± 1.42 μm/min; N = 40 microtubules from 12 cells; Figures 3a,b,e; Table 1), cortical microtubules of ktn1-2 mutant show significantly slower plus end growth kinetics (5.6 ± 1.59 μm/min; N = 36 microtubules from 14 cells; p < 0.001; Figures 3c–e; Table 1). Other measures of microtubule dynamics, including plus end shrinkage (18.7 ± 3.14 μm/min for Col-0; N = 39 microtubules from 12 cells vs. 17.47 ± 3.19 μm/min for ktn1-2; N = 40 microtubules from 14 cells; Figure 3f; Table 1), minus end growth (1.12 ± 0.4 μm/min for Col-0; N = 18 microtubules from 12 cells vs. 1.12 ± 0.47 μm/min for ktn1-2; N = 14 microtubules from 8 cells; Figure 3g; Table 1), and minus end shrinkage (1.39 ± 0.42 μm/min for Col-0; N = 26 microtubules from 20 cells vs. 1.37 ± 0.4 μm/min for ktn1-2; N = 25 microtubules from 22 cells; Figure 3h; Table 1) revealed no significant differences. It was also noteworthy that ktn1-2 mutants showed significantly lower catastrophe and rescue frequencies (0.017 ± 0.003 and 0.011 ± 0.004 events/s, respectively from a total of 40 microtubules compared to Col-0, 0.022 ± 0.006 and 0.016 ± 0.008 events/s, respectively from a total of 39 microtubules; Figures 3i,j; Table 1).

As expected, ktn1-2 mutants exhibit completely abolished severing activity resulting in the absence of severing in expected sites of the cortical array such as microtubule crossovers (Figures 4a–c; Table 1). In particular, the severing frequency in Col-0 petiole epidermal cells was 0.145 ± 0.0918 events × 10⁻³ × μm⁻² (N = 19 cells) while in ktn1-2 petiole epidermal cells the respective frequency was 0 (N = 16 cells). The absence of severing in ktn1-2 mutants was most striking at predominant severing sites such as microtubule crossovers (e.g., Figure 4b).

Major defects in microtubule organization of ktn1-2 mutant were observed in the case of PPB formation and narrowing as studied comparatively in Col-0 and ktn1-2 seedlings stably transformed with a GFP-TUA6 microtubule marker. Col-0 cells exhibit a well-organized PPB comprising of parallel and straight microtubules (Figure 5a) with perpendicular orientation (Figure 5b). By contrast, ktn1-2 preprophase cells mostly show very broad and often splayed PPBs with asymmetric organization containing over elongated and frequently bent microtubules localized outside of the PPB region (Figures 5c1,c2,d; Video S1) as well as fan-shaped, broad, disordered, and an incomplete PPBs (Figures 5e–g). It is also worth to note that many such PPBs of the ktn1-2 mutant were incomplete as deemed by 3D rendering, failing to circumvent the entire cell cortex.

In the 3 dimensions, the PPB of Col-0 (Figures 6a–d3; Video S2) encompasses the entire cortical circumference with a rather uniform microtubule band. By contrast, PPBs of ktn1-2 are non-uniform in width while sometimes they are incomplete failing to form a complete ring (Figures 6e–g3; Video S3). In quantitative terms, all PPBs examined in Col-0 (over 50 PPBs) were normal, while PPBs of ktn1-2 were variably abnormal. In this case we discriminated 3 distinct categories: those with asymmetry of their organization extent (one side well organized and the other poorly organized; category I with 21 cases in total; Figures 5c1,c2,e1,e2), those which exhibited a roughly uniform width, however, with residual long cortical microtubules outside the PPB area (category II with 12 cases in total; Figure 5f) and those which were incomplete (category III with 8 cases in total; Figures 5g, 6g1–g3).

Time lapsed imaging and measurement of the PPB narrowing process showed that by contrast to PPB of Col-0 which narrows discernibly in a relatively short time (up to 1 h; Figures 7a,c,d; Video S4), the PPB of ktn1-2 cells narrows at a much slower pace (sometimes exceeding 3 h) resulting in a broader PPB compared to Col-0 (Figures 7b–d; Video S5). Owing to the very low laser power used during documentation (0.07 mW at the focal plane of a 63 × 1.40 NA oil immersion objective) and the fact that such cells successfully enter and complete mitosis and cytokinesis (Videos S4, S6), phototoxicity can be excluded as the cause for the prolonged persistence of the PPB in ktn1-2. The time lapsed series, showed that sometimes, the different forms of PPB described above (Figure 5) maybe interchangeable to a certain extent. In this respect the cell shown in Figure 7b, Videos S5, S6, starts from category II (panels 0–90 min) and transforms to a reminiscent of category I (panels 100–210 min) although not as exaggerated as shown in Figures 5e1,e2.

Mitotic spindle assembles normally in ktn1-2 mutants as compared to Col-0 and assumes bipolarity quite early during prophase (Figures 8a,b). In all cases examined we never observed aberrations in mitotic spindle polarity as previously published (Panteris et al., 2011), however, we corroborated the residence of cortical microtubules reminiscent of PPB in late mitotic cells (Figures 8b1–b3; Video S7; see also Panteris et al., 2011). Likewise, the overall architecture of phragmoplast is conserved between Col-0 (Figure 8c) and ktn1-2 mutants (Figures 8d–g), showing frequently the previously described “arrowhead” configuration, especially at late stages of phragmoplast expansion (Figures 8e–g).

Following the kinetics of the mitotic and cytokinetic processes and compared to Col-0 (Figures 9a,d; Video S8) mitotic progression is also slightly but significantly delayed in ktn1-2 (Figures 9b,d; Video S6). In contrast to Col-0 (Figures 9a,e; Video S8) the course of centrifugal phragmoplast expansion is more appreciably delayed in ktn1-2 (Figures 9b,c,e; Videos S6, S9). Although, the expanding phragmoplast of ktn1-2 has to frequently span longer wall-to-wall distances compared to Col-0 (Figure 9f), it does so at considerably lower rates (Figure 9g). It is notable however, that albeit the prolongation of phragmoplast expansion, we never observed abortive cytokinesis and the phragmoplast always expanded until it met the parent wall (see next section).

In Col-0 cells, mitotic spindle assembles in such way that its equatorial plane coincides with the plane set by the PPB. Previous studies have shown that at least in the case of the root, KATANIN 1 deficiency in either erh3 or in fra2 and lue1 mutants (Webb et al., 2002; Panteris et al., 2011) results in disturbances of cell division plane orientation.

In dividing cells of Col-0, cell division plane is predetermined by the PPB. During mitosis the equatorial plane of the mitotic spindle coincides with the PPB plane and finally, it is followed by the expanding phragmoplast (Figure 10a). In dividing ktn1-2 cells, the mitotic spindle assembles at a regular bipolar form (Figure 10b). However, it exhibits random rotational motions so that the equatorial plane deviates significantly from the PPB plane. After mitosis is completed and the phragmoplast is formed, it undergoes similar motions until one end meets the parent wall at the PPB site (Figure 10b, red outlined frames). As soon as one of the phragmoplast ends is stabilized in this manner the rest cytokinetic process is rectified so that the phragmoplast plane is expanding at the plane set by the PPB (Figure 10b, panels 80 and 96 min).

Katanin is a master regulator of cortical microtubule organization (Nakamura, 2015) having essential roles in biasing their parallel arrangement during cell elongation. Previous studies showed that KATANIN 1 severs nascent microtubules formed by branched formation on preexisting microtubule walls (Nakamura et al., 2010) and becomes selectively activated at microtubule crossovers (Wightman and Turner, 2007; Wightman et al., 2013; Zhang et al., 2013). During the reorganization of cortical microtubules in response of blue light induction it was shown that a mechanism responsible for the selective activation of KATANIN 1 at microtubule intersections is via the activation of the PHOT1 and PHOT2 phototropin photoreceptors (Lindeboom et al., 2013). Additionally, KATANIN 1 was shown to regulate the mechanical response of cortical microtubule reorganization and establishment of anisotropy during the growth of the shoot apical meristem (Uyttewaal et al., 2012) by antagonizing the action of auxin (Sassi et al., 2014). Moreover, KATANIN 1 activity is regulating cytomorphogenesis of epidermal pavement cells, downstream of the small GTPase ROP6 (Lin et al., 2013).

In the present study, we quantified organizational defects of cortical microtubules in ktn1-2 mutants when compared to Col-0. As parameters we used angular distribution which reflects to the ordering capacity of the cortical array, skewness which is a measure of microtubule bundling. Such correlative study was quite important since the severing activity of KATANIN 1 in the above cells is strikingly different, especially regarding severing at microtubule crossroads. In this case it seems, that such events are much more frequent in petiole than in cotyledon epidermal cells (Wightman et al., 2013).

As schematically shown in Figure 11, the failure to sever microtubules at various situations depicted, can explain misorganization of cortical microtubules, since KATANIN 1 plays a key role to symmetry breaking and to anisotropy induction in the cortical array during polar or directed cell growth by biasing the parallel organization of cortical microtubules (Wightman and Turner, 2007, 2008; Nakamura et al., 2010; Uyttewaal et al., 2012; Lindeboom et al., 2013; Zhang et al., 2013; Chen et al., 2014; Sassi et al., 2014).

Within studies addressing the role of KATANIN 1 in cortical microtubules its impact on the dynamics of these microtubules was not addressed. In the present study we found that most measures of microtubule dynamics in ktn1-2 cells were comparable to those of Col-0 with the exception of the plus-end growth rate (which was considerably reduced). Meanwhile the shrinkage rate was only slightly affected, but importantly the overall catastrophe frequency was also considerably reduced. As was shown by previous study on KATANIN 1 overexpression, it seems that KATANIN 1-mediated microtubule severing somehow favors microtubule bundling (Stoppin-Mellet et al., 2006). Therefore, the considerable reduction of microtubule bundling as judged by reduction in fluorescence skewness in ktn1-2 petiole and cotyledon epidermal cells was expected. Microtubule bundling is considered as a major mechanism of ordered cortical microtubule organization mediated by members of the MAP65 protein family (Lucas et al., 2011; Lucas and Shaw, 2012). For this reason we propose that the effect of KATANIN 1 in cell growth directionality is not only direct by promoting the disassembly of microtubules at unfavorable orientations but also indirect by regulating microtubule bundling. These two discrete functions may explain cortical microtubule disordering and isotropic cell growth in the ktn1-2 mutant. When katanin severs a microtubule, it promotes the exposure of GDP-tubulin ends which will inevitably undergo catastrophic shrinkage (Mace and Wang, 2015).

The present study is the first one to provide mechanistic insight on the involvement of KATANIN 1 in mitotic and cytokinetic progression of Arabidopsis cells through time lapsed imaging. The PPB is formed at the S to G2 interphase of the cell cycle by the large scale rearrangement of cortical microtubules to a broad annular microtubule assembly that progressively narrows until it finally disappears at the onset of mitosis concomitantly with nuclear envelope breakdown (Vos et al., 2004; Marcus et al., 2005; Azimzadeh et al., 2008). The duration of PPB persistence from its formation until its disappearance may be prolonged (e.g., over 11/2–2 h as recorded in tobacco BY-2 cells; Dhonukshe and Gadella, 2003; Vos et al., 2004) while in cases it may considerably exceed the duration of the entire mitotic procedure (as estimated for Azolla pinnata; Gunning et al., 1978). In our case, delays in PPB maturation/narrowing were inferred by comparison of the same cell types (i.e., dividing cotyledon pavement cells and dividing petiole epidermal cells). Of its known roles, the PPB accurately predicts the insertion of the succeeding cell plate during cytokinesis (Pickett-Heaps and Northcote, 1966; Rasmussen et al., 2011 and references therein) and represents a cortical site of intense endocytotic uptake (Karahara et al., 2009). Very interesting recent study proposed that PPB regulates robustness of cell division orientation by limiting spindle rotations (Schaefer et al., 2017). In this respect, we have found spatio-temporaly disturbed PPBs along with vigorous spindle and phragmoplast rotations in the ktn1-2 mutant. They suggest KATANIN 1 role in the positioning of PPB, spindle, and phragmoplast.

Early steps of microtubule bundling during the onset of PPB formation are mediated by short actin microfilaments (Takeuchi et al., 2016) while the occurence of actin within the PPB is considered to be crucial for its narrowing which is prevented upon actin depolymerization (Granger and Cyr, 2001). The inability of ktn1-2 preprophase cells to form a regular and progressively narrowing PPB, implicates KATANIN 1 microtubule severing activity in controlling microtubule assembly during both processes. Therefore, it is not surprising that ectopic microtubules, i.e., microtubules that reach cortical areas outside the PPB zone, seem to emanate from the PPB.

The role of PPB in the determination of cell division plane orientation has been extensively documented through the localization of cortical marker proteins such as TANGLED, FASS, PHRAGMOPLAST ORIENTING KINESINS 1 and 2, AIR9, and RanGAP1, the localization of which either coincides or even persists at the cortical PPB site throughout mitosis and cytokinesis (Camilleri et al., 2002; Buschmann et al., 2006; Müller et al., 2006; Walker et al., 2007; Xu et al., 2008). Beyond the localization of the above protein markers at the PPB site and the defects of cell division plane orientation observed in Arabidopsis mutants of the above proteins, the mechanisms attracting the phragmoplast at the cortical cell division plane and guiding cell plate deposition coincidentally to the PPB plane remain largely elusive.

In accordance to previous localization of KATANIN 1 within the PPB (Panteris et al., 2011), we postulate that KATANIN 1 is not responsible for clearing the rest of the cell cortex from residual microtubules but rather that KATANIN 1 activity is restricted within the PPB to regulate microtubule length and in this way KATANIN 1 depletion in ktn1-2 perturbs PPB shape, organization and position. From our kinetic analysis, it became also evident that PPB narrowing is considerably prolonged, further suggesting that KATANIN 1 activity is also implicated in PPB maturation.

In such wide PPB areas defined in ktn1-2 mutants, it is hypothetically possible that markers of CDP orientation attracting cell plate deposition to the PPB site (Lipka et al., 2015) may be more diffusely localized and this would help to explain the occurence of oblique cell division planes observed in KATANIN 1 mutants such as erh3, fra2, lue1, and ktn1-2 (e.g., Webb et al., 2002; Panteris et al., 2011). In living mitotic petiole and cotyledon epidermal cells of ktn1-2 we never observed multipolar spindles as those described before by tubulin immunofluorescence in fixed root whole mounts of fra2 and lue1 (Panteris et al., 2011). Mitotic spindles of ktn1-2 were uniformly bipolar and identical in shape compared to those of Col-0. What differed remarkably between the two samples (Col-0 and ktn1-2) was that mitotic spindles of ktn1-2 exhibited large scale rotational motions with the equatorial plane deviating considerably from the CDP as it is defined by the PPB. After the completion of mitosis, the young phragmoplast is also very motile showing a propensity to co-align with the PPB site. Once one part of the phragmoplast is stably anchored to the parent wall, it does so on the PPB site while the remaining of phragmoplast margin moves along the PPB plane until it meets the opposite PPB cortical site. By contrast to what was published before (Panteris et al., 2011) spindle polarity, form, or position, is not related to CDP orientation since cell plate deposition uniformly follows the PPB plane in ktn1-2 irrespectively of spindle orientation. Therefore, whenever CDP orientation defects are observed in KATANIN 1 mutants (Webb et al., 2002; Panteris and Adamakis, 2012) they can be attributed to PPB malorganization or misorientation.

In ktn1-2 mutant most of the PPBs documented show an asymmetry with respect to their organization degree with one very well focused and one poorly and very broad side. In this case, it might be that the expanding phragmoplast might establish firm contact with the cortical site corresponding to the well-organized PPB side, however, it would have problems to navigate toward the ill-defined side.

In dividing epidermal cells of the petiole or the cotyledon of ktn1-2 mutant, the spindle forms normally prior to PPB breakdown. As a matter of fact, the mitotic spindle may progress to metaphase in the presence of prominent residual PPB-like cortical microtubule structure, suggesting that the complete disassembly of the PPB is not a prerequisite for normal mitotic spindle assembly.

The mitotic spindle of the ktn1-2 mutant at any stage of its organization (prophase, metaphase, and anaphase) does not seem to differ morphologically from its Col-0 counterpart, further implying that KATANIN 1 microtubule severing is not required to guide spindle assembly and transitions. To this extend we never observed the previously reported spindle multipolarity during prophase as reported before (Panteris et al., 2011; Panteris and Adamakis, 2012). Moreover, and in contradiction to animal cells, A. thaliana KATANIN 1, does not seem to affect mitotic spindle size as was shown through comparison between two different Xenopous species (Loughlin et al., 2011).

In contrast to Col-0 where the spindle showed minimal movement in the cytoplasmic space, ktn1-2 mitotic spindles at any stage displayed broad rotational motions such that the equatorial plane significantly deviated from the PPB plane. This observation suggests that KATANIN 1 activity is somehow required for spindle positional control, possibly by controlling attachment of the spindle to the cell cortex (Granger and Cyr, 2001; Kojo et al., 2013, 2014; Lipka et al., 2015). Similar motions were also exhibited by the nascent phragmoplast at the earliest stages of its formation and before its margin could become tethered to the parent wall. Nevertheless, the motions of the phragmoplast can be considered as corrective as they always concluded with the attachment of its margin to predetermined cortical sites. Therefore, anchoring of the phragmoplast to the “correct” cortical sites (i.e., those determined by the PPB) is not prevented in the ktn1-2 mutant.

KATANIN 1 depletion also interferes with the course of mitosis but most remarkably with the duration of cytokinesis, although it does not prevent the complete cell plate deposition, neither does it uncouple its positional relationship with the PPB. The centrifugal expansion of the phragmoplast requires a continuous assembly of microtubules at its margin followed by microtubule disassembly at the lagging parts toward the center of the cell plate (Murata et al., 2013). Similar delays in phragmoplast expansion were observed in the Arabidopsis signaling mutant mpk4, lacking the activity of the mitogen activated protein kinase 4 (Beck et al., 2011). Such results indicate that the coordinated microtubule unbundling via MPK4-dependent phosphorylation of members of the MAP65 microtubule bundling proteins (Beck et al., 2010; Sasabe et al., 2011) and KATANIN 1-mediated microtubule severing, may facilitate the clearing of microtubules from the lagging areas of the phragmoplast and allow its centrifugal expansion solely at the margin.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.