Quantitative analysis of changes in actin microfilament contribution to cell plate development in plant cytokinesis

Background Plant cells divide by the formation of new cross walls, known as cell plates, from the center to periphery of each dividing cell. Formation of the cell plate occurs in the phragmoplast, a complex structure composed of membranes, microtubules (MTs) and actin microfilaments (MFs). Disruption of phragmoplast MTs was previously found to completely inhibit cell plate formation and expansion, indicative of their crucial role in the transport of cell plate membranes and materials. In contrast, disruption of MFs only delays cell plate expansion but does not completely inhibit cell plate formation. Despite such findings, the significance and molecular mechanisms of MTs and MFs remain largely unknown. Results Time-sequential changes in MF-distribution were monitored by live imaging of tobacco BY-2 cells stably expressing the GFP-actin binding domain 2 (GFP-ABD2) fusion protein, which vitally co-stained with the endocytic tracer, FM4-64, that labels the cell plate. During cytokinesis, MFs accumulated near the newly-separated daughter nuclei towards the emerging cell plate, and subsequently approached the expanding cell plate edges. Treatment with an actin polymerization inhibitor caused a decrease in the cell plate expansion rate, which was quantified using time-lapse imaging and regression analysis. Our results demonstrated time-sequential changes in the contribution of MFs to cell plate expansion; MF-disruption caused about a 10% decrease in the cell plate expansion rate at the early phase of cytokinesis, but about 25% at the late phase. MF-disruption also caused malformation of the emerging cell plate at the early phase, indicative of MF involvement in early cell plate formation and expansion. The dynamic movement of endosomes around the cell plate was also inhibited by treatment with an actin polymerization inhibitor and a myosin ATPase inhibitor, respectively. Furthermore, time-lapse imaging of the endoplasmic reticulum (ER) revealed that MFs were involved in ER accumulation in the phragmoplast at the late phase. Conclusion By expression of GFP-ABD2 and vital staining with FM4-64, the dynamics of MFs and the cell plate could be followed throughout plant cytokinesis in living cells. Pharmacological treatment and live imaging analysis also allowed us to quantify MF contribution to cell plate expansion during cytokinesis. Our results suggest that MFs play significant roles in cell plate formation and expansion via regulation of endomembrane dynamics.


Background
Eukaryotic cell division is required for the separation of duplicated chromosomes and the subsequent step of physical division of the mother cells into two daughter cells. In plants, the process of chromosomal separation is generally similar to that of animal cells [1,2], whereas the mechanism of physical cell division, called cytokinesis, differs strikingly between plants and animals. In plants, the cell membrane-enveloped juvenile cell wall, referred to as the cell plate, accumulates between the daughter nuclei in late anaphase, and the cell plate subsequently expands centrifugally so as to divide the parental cell into two [3,4]. Cell plate formation and expansion are performed by structures that sandwich the developing cell plate, known as the phragmoplast, which is composed of membranes and cytoskeletons.
The cytokinetic membranes are generally thought to originate from Golgi-derived vesicles [3,[5][6][7]. During cytokinesis, Golgi stacks were seen to accumulate and to become associated with the cell plate in living tobacco BY-2 cells [8] and in cryopreserved shoot apical meristematic cells of Arabidopsis thaliana [9]. It was recently proposed that endocytic vesicles also participate in cell plate formation [10], and this was based on the localization of several cell wall and/or plasma membrane (PM) materials in the cell plate; including xyloglucans [11,12], pectin [10,12], PM proteins, such as the auxin transport facilitators, PIN1 [13] and PIN2, the water channel, PIP2, the integral PM protein, LTI6b, and the brassinolide receptor, BRI1 [10], dynamin [10], clathrin [14], and the endosomal marker proteins, GNOM-Myc, Ara6 and Ara7 [10]. However, a more recent study demonstrated that endocytosis is not essential for cell plate formation [7]. The role of endocytosis in cell plate formation thus remains controversial.
Microtubules (MTs) are a major component of the phragmoplast, and the MTs and predicted kinesin motor proteins play a crucial role in the transport of cell plate materials [15,16]. The plus-ends of MTs face the cell plate [17] and terminate within the cell plate assembly matrix [6]. Recent electron tomographic analysis has revealed, with nano-scale resolution, that the three-dimensional geometry of phragmoplast MTs undergoes dynamic changes [18].
Actin microfilaments (MFs) are also distributed throughout the phragmoplast and are aligned parallel to the MTs, as reported in endosperm cells of the blood lily, Haemanthus katherinae [19][20][21][22], and the kaffir lily, Clivia nobilis [23], in root cells of onion, Allium cepa [24,25], in stamen hair cells of spiderwort plants, Tradescantia virginia [26][27][28], in cultured cells of carrot, Daucus carota [29], and alfalfa, Medicago sativa [30], as well as in tobacco BY-2 cells [31,32]. Electron microscopy-based, nano-scale structures of phragmoplast-related MFs have not yet been reported, mainly because of technical limitations [6]. Numerous studies have shown that, unlike MT inhibitors, actin inhibitors could not inhibit the initiation of the cell plate [27,[33][34][35][36][37]. However, an actin polymerization inhibitor was found to delay cell plate expansion in Haemanthus katherinae endosperm cells [20]. Microinjection of profilin, a monomeric actin binding protein, into dividing Tradescantia stamen hair cells was found to disrupt phragmoplast-associated MFs and to inhibit cell plate expansion [27]. Furthermore, treatment with 2,3butanedione monoxime (BDM), an inhibitor of myosin ATPase, or with ML-7, a specific inhibitor of myosin lightchain kinase, could delay the complete fusion of the cell plate with the parental cell wall [38]. Based on these results, MFs have been postulated to contribute to cell plate expansion at the late phase of cytokinesis. However, the mode of MF contribution to plant cytokinesis still remains unclear.
In this study, we have quantitatively estimated the timesequential changes in MF contribution to cell plate development using live cell image analysis of tobacco BY-2 cell lines. In addition, we examined the relationship between MFs and endomembrane dynamics, including the endosomes and endoplasmic reticulum (ER). Our results detail new features of MFs in plant cytokinesis via regulation of the endomembrane system.

Dynamic changes in MF localization during plant cytokinesis
To simultaneously observe MFs and the cell plate, tobacco BY-GF11 cells, in which MFs were visualized by expression of a GFP-ABD2 fusion protein [37], were stained with the styryl dye, FM4-64. After pulse-labeling with this endocytic tracer, the fluorescence moves from the PM to endosomes, and finally reaches the vacuolar membranes [39][40][41]. In addition, FM4-64 localizes to the cell plate membranes of Arabidopsis root cells [10,42] and tobacco BY-2 cells [40,43], in agreement with the previously suggested involvement of the endocytic pathway in membrane trafficking to the cell plate [10].
From 5 min to 3 h after pulse-labeling with 16 μM FM4-64, we observed that the fluorescence became localized mainly in the cell plate membranes and endosomes of the cytokinetic BY-GF11 cells (Fig. 1A), similar to that of wild type tobacco BY-2 cells (data not shown). At this time, there was still a slight remnant of fluorescence in the PM but not in the vacuolar membrane. For cell plate observations, all results presented in this report were, unless otherwise indicated, from those conducted 2 h after pulselabeling.
At metaphase, no prominent MFs or FM4-64 fluorescence was observed in the mitotic apparatus (Fig. 1A, 0 min), whereas cortical MF twin peaks [37] were evident (Fig. 1A, 0 min, asterisks). From anaphase to telophase, the FM4-64 fluorescence gradually accumulated at the equatorial region, and the cell plate subsequently emerged (Fig. 1A, 4-6 min) as previously reported [10]. Concomitantly, there was a gradual accumulation of MFs from the periphery of the daughter nuclei towards the emerging cell plate (Fig. 1A, 4-6 min). At telophase, the MFs became localized near the expanding cell plate (Fig. 1A, 8-13 min). Measurements of fluorescent intensity around the division plane illustrate the accumulation of MFs in parallel with development of the FM4-64-labelled cell plate (Fig.  1B). To evaluate the spatial changes in MFs, a kymograph was constructed by tracking the fluorescence across the developing cell plate (Fig. 1C). The kymograph suggests that the MFs gradually approached the cell plate (Fig. 1C, yellow broken lines).
To examine the structural relationship between MFs and cell plate during cytokinesis, the GFP-ABD2 and FM4-64 fluorescent intensities were measured along the cell plate at early ( Fig. 2A, line segment PQ) and late (Fig. 2F, line segment RS) telophase. GFP-ABD2 was detected all along the equatorial plane, but its intensity was lower than that of FM4-64 in the cell plate at early telophase (Fig. 2B). In contrast, the intensity of GFP-ABD2 was high, especially at the edges of the cell plate, at late telophase (Fig. 2G). These results indicate that only a few MFs were localized Time-lapse imaging of actin microfilaments and cell plate in the mitotic apparatus at early telophase, but that they accumulated around the expanding cell plate at late telophase. A three-dimensional cross-sectional image, reconstructed from serial optical sections, confirmed these localization patterns at early (Fig. 2C, D and 2E) and late (Fig. 2H, I and 2J) telophase. These results suggest that MFs undergo dynamic changes in their arrangement during cell plate development.

Quantitative analysis of the effects of an actin polymerization inhibitor on cell plate expansion
As mentioned above, vital staining with FM4-64 allowed us to follow chromosomal separation (Fig. 3A, -1-0 min), the gradual accumulation of cell plate vesicles (Fig. 3A, 1-2 min), cell plate expansion (Fig. 3A, 3-9 min), and final division of the parental cell (Fig. 3A, 12 min). To study the role of MFs in cell plate development, these sequential events were monitored under MF-disrupted conditions induced by pre-treatment with the dimeric macrolide, bistheonellide A (BA), an inhibitor of actin polymerization [44]. Treatment with 1 μM BA for 1 h almost completely destroyed the MF networks and resulted in the dispersion of free GFP-ABD2 fluorescence in the cytoplasm (Fig. 3B). A 1-h pre-treatment with 1 μM BA caused a temporary malformation of the emerging cell plate, which thus appeared to be discontinuous (as described further below; Fig. 3C, 2 min), and also increased the time needed for the expanding cell plate to complete its fusion with the parental cell wall independently of cell size (Fig.  3D). These results suggest that MFs partially contribute to cell plate formation and expansion.
To study the time-dependency of MF contribution to cell plate expansion, we sequentially measured the cell plate diameters from time-lapse images. Measurements of cell plate diameter and estimations of cell plate area showed that the expansion rate decreased with time in BA-treated cells ( Table 1, BA treatment), whereas the rate was roughly constant in control cells (Table 1, control). These results suggest that the contribution of MFs to cell plate expansion increases with the progress of cytokinesis.
To quantitatively estimate this contribution of MFs, we used a simple mathematical model for cell plate expan-Three-dimensional localization of actin microfilaments at early and late telophase  sion, based on the well-known concept of a constant increase in cell plate area [45]. In turn, the diameter increased with the square root of time, as follows:

Effects of bistheonellide A (BA) on the duration required to complete cytokinesis
where D is the cell plate diameter, a is the expansion rate, and t is time. The time-lapse data of the diameter fit equation 1 well for the control cells ( However, to compare the contributions of MFs to cell plate expansion in control and BA-treated cells, both data sets need to be applied to the same model equation. When we applied time-lapse data from control cells to model equation 2, n became underspecified because b came close to 0 (data not shown). We, therefore, assumed that n is invariant at 1.81, even in the control cells, and applied the control data to model equation 2 (Fig. 4B,  control). Consequently, the control data precisely fit model 2 as well as model 1 (Fig. 4C). The mean values of a and b were 7.61 and 0.006, respectively (Fig. 4B, control). Simulating the expansion rate of the cell plate with model equation 2 quantitatively estimated the changes in the expansion rate (i.e. a-bt n in model equation 2) during cytokinesis in control and BA-treated cells (Fig. 4D). Finally, the contribution of MFs to cell plate expansion could be time-sequentially estimated by the differences between expansion rates in control and BA-treated cells (Fig. 4E). The contribution of MFs to cell plate expansion was found to be about 10% just after cell plate formation and gradually increased up to about 25% at the end of cytokinesis (Fig. 4E). ND: Not defined. The diameter of the FM4-64-stained cell plate was measured as shown in Figure 4A. Values are arithmetic means ± SE, n = 12. Cell plate area was estimated by assuming that the cell plate is a perfect circle (Area = π (Diameter) 2 /4). Diameters at 1 and 2 min after initiation of cell plate formation (Table 1, BA treatment) could not be defined because the emerging cell plate was deformed in BA-treated cells (see Fig. 3C, 2 min and Fig. 5).

MF-disruption-induced a temporary malformation of the emerging cell plate
Interestingly, a 1 h pre-treatment with 1 μM BA caused malformation of the emerging cell plate, which thus appeared discontinuous (Fig. 3C, 2 min). This aberration was only temporary, in that it could be observed for a maximum of only 1-2 min (Fig. 3C, 2 min), after which the blank areas of the cell plate were filled in so that it attained a normal appearance (Fig. 3C, 3 eral holes (Fig. 5A, upper panels). To assess their morphology, we extracted the cell plate area from the crosssections (Fig. 5A, lower panels), and measured their area, perimeter and complexity using ImageJ software [46]. Complexity was defined as a morphological parameter and calculable from the area and perimeter (see Methods) [47]. A circle has the lowest value, 1.0, of complexity and, as the structure becomes more complicated, the complexity reaches appropriately higher values [47]. The mean value of complexity of cell plates in BA-treated cells was higher than that of control cells (Fig. 5B), whereas their mean areas did not differ significantly (Fig. 5C). In addition, aniline blue staining revealed that the deformed cell plate contained callose (Fig. 5D) as in the control cells (data not shown). These results suggest that MFs modulated cell plate formation/expansion at the early phase of cytokinesis, but did not affect callose synthesis.

Actin-myosin-dependent endosomal movement around the cell plate To investigate the interactive mechanisms of MFs and cell plate formation/expansion, the relationship between MFs
Effects of bistheonellide A (BA) on morphology of emerging cell plates and endosomes, which have been implicated in cell plate development [10], was examined. In a control cell at the late phase, there was active movement of numerous endosomes ( Fig. 6C and 6F) around the edge of the expanding cell plate (Fig. 6A). Some endosomes were found to interact with the edge of the expanding cell plate (Fig. 6B). To examine the role of actin-myosin systems on endosomal movement around the cell plate, we treated the cells with BA or 2,3-butanedion monoxime (BDM), a general myosin ATPase inhibitor [48,49]. The movements were discernibly inhibited by BA (Fig. 6D and 6G) and BDM pre-treatments ( Fig. 6E and 6H). These inhibitory effects were confirmed statistically by measuring the velocity of the endosomal movements (Fig. 6I, J and 6K). Dual observations of MFs and endosomes showed that the endosomes moved towards the cell plate along the MFs (Fig.  6L, M and 6N). These results suggest that endosomal movement around the edge of the expanding cell plate depends on the actin-myosin systems.

MF-dependent accumulation of ER in the phragmoplast
To further test the involvement of MFs in endomembrane dynamics during cytokinesis, we analyzed the effects of BA treatment on ER-organization during cytokinesis in tobacco BY-2 cells expressing SP-GFP-HDEL, which is retained within the ER [50]. ER accumulated in the spindle poles in meta-/anaphase (Fig. 7A, 0-3 min), sandwiched the cell plate in telophase (Fig. 7A, 6-21 min), and accumulated in the phragmoplast in late telophase (Fig. 7A, 18-21 min, yellow arrowheads), as previously reported [6,8,51]. A 1 h BA pre-treatment did not induce marked changes in ER-organization at metaphase (Fig.  7C, 0 min), but did inhibit ER-accumulation in the phragmoplast at late telophase (Fig. 7C, 48-63 min, yellow arrows). Three-dimensional images also showed a significant decrease in the amount of ER in the phragmoplast ( Fig. 7B and 7D). In addition, time-sequential measurements of fluorescent intensity around the division plane illustrated the inhibition of ER-accumulation in the phragmoplast (Fig. 7E).

Discussion
In this study, we have quantitatively examined the temporary changes in MF contribution to cell plate expansion. To visualize the MFs and cell plate, we used GFP-ABD2 expression and FM4-64 vital staining, respectively. Although GFP-ABD2 is recognized as a more reliable MFmarker than GFP-mouse talin (GFP-mTn) [37,[52][53][54], recent studies have shown that overexpression of GFP-ABD2 can cause growth inhibition of pollen tubes [55] and organelle motility [56], suggesting the need for care and attention in the use of such GFP markers. Our BY-GF11 cell line shows a normal growth rate compared with wild-type BY-2 cells (data not shown) and appears to have minimal side effects. Furthermore, FM4-64, which has been widely used as an endocytosis marker, was found to induce abnormal membrane systems in fungal cells [57]. However, FM4-64 cytotoxicity appears limited in plant cells [40] and, at least under our experimental conditions [39], abnormal membrane structures or an inhibition of cell division in the BY-2 cells were not observed. These labeling techniques, which cause minimal cell damage, have allowed us to study MF-dynamics and their roles in cell plate development.

MFs contribute to cell plate expansion in late expansion phase, but even in early initiation phase
It is widely accepted that there are two stages in plant cytokinesis that can be distinguished by cell morphological features and sensitivity to caffeine [28]. The first stage is the early initiation phase, which starts immediately after chromosomal separation and involves the deposition and early expansion of the cell plate until it reaches the width of the daughter nuclei. The second stage is the late expansion phase that involves growth of the cell plate beyond the width of the nuclei until it reaches and fuses with the parental cell wall. The late expansion phase is completely inhibited by caffeine [28], which also inhibits callose deposition and the organization of phragmoplast MTs [58].
MFs are thought to be also involved in the late expansion phase of the cell plate. In cytochachalasin-treated Haemanthus katherinae endosperm cells, the cell plate often remains incomplete [20]. Similar cell plate inhibition has been reported in profilin-microinjected Tradescantia virginiana stamen hair cells [28], and in Tradescantia stamen hair cells treated with myosin inhibitors [38]. In this study, the increase in MF contribution to the late expansion phase of the cell plate, which was confirmed by timesequential measurements and data fitting analysis (Fig. 4), concurs with previous reports and also with the gradual accumulation of MFs near the cell plate found in this study (Fig. 1C). On the other hand, our results revealed that MF-disruption decreased the expansion rate by about 10%, even just after cell plate formation, but by about 25% at the end of cytokinesis (Fig. 4E) and, in addition, caused malformation of the emerging cell plate (Fig. 5). These results suggest that MFs not only facilitate cell plate expansion in the late expansion phase but also in the early initiation phase.
Furthermore, MF-disruption-induced malformation of the cell plate suggests that MFs regulate the early disc-like morphology of the cell plate. In the early phase, the MFs had not yet accumulated near the cell plate, but a thin, widely-spread MF-meshwork could be observed in the mitotic apparatus and near the cell plate (Fig. 1A, Fig. 2A-E). The existence of MFs near the cell plate in the early phase was also observed by rhodamine-phalloidin staining of fixed cells of onion Allium cepa [24]. BA-induced cell plate malformation is reminiscent of the mini-phragmoplasts, which appear during Arabidopsis endosperm cellularization [59,60] and in post-meiotic cytokinesis during pollen development [61]. Mini-phragmoplasts consist of 2 to12 overlapping MT clusters, assembled between duplicated nuclei, and spaced on average 0.6 μm apart from each other across a given division plane [61]. BA-induced cell plate blank regions were approximately of the same size (Fig. 5A, BA treatment). The fact that similar structures occur by MF-disruption suggests that MFs are involved in the spread and flattening of the cell plate at the early phase.

Actin-myosin system is possibly involved in plant cytokinesis via endosomal dynamics
In this study, we found MF-disruption had parallel, aberrant effects on cell plate development and endosomal movement around the expanding cell plate (Fig. 6D, G and 6J). Treatment with the general myosin inhibitor, BDM, revealed that endosomal movements depended on the actin-myosin system (Fig. 6E, H and 6K). MF-dependent endosomal dynamics were also reported in root cells of Arabidopsis thaliana [13,[62][63][64], Zea mays [65,66], Medicago truncatula [67] and pollen tubes of Picea meyeri [68]. Endosome association with MFs was observed in cytokinetic BY-2 cells (Fig. 6L, M and 6N) as well as in root hair cells [67]. From these results, we hypothesize that the actin-myosin systems are partially involved in endocytic membrane traffic to the cell plate. This notion is supported by recent studies demonstrating that the plant myosin class VIII, ATM1, is localized in the expanding cell plate of tobacco BY-2 cells [69] and in FM4-64-labelled endosomes of Nicotiana benthamiana leaves [70]. Although the endocytic pathway might participate in membrane trafficking to the cell plate [10], it is not essential for completion of cytokinesis [7]. Indeed, as Golgiderived vesicles transported by MTs and unidentified kinesins are necessary and sufficient for cell plate development [7,15], the contribution of MFs to cell plate formation and expansion may be rather limited. Unfortunately, we could not examine the effects of BDM on cell plate expansion, because BDM-pretreatment inhibited cell division in the BY-2 cells (data not shown), possibly through their side effects on MTs [48]. Future studies, using combinations of genetic inhibition of myosin activity and time-sequential morphometry as shown in this study, will reveal the changes in contribution of myosins to cell plate expansion.

MFs regulate ER-accumulation in the phragmoplast at late expansion phase
The ER is one of the first organelles to be visualized during plant cytokinesis. Numerous electron microscopic studies have shown that the tubular ER is first arranged parallel to the spindle MTs in metaphase, and then gradually accu-mulates near the cell plate and phragmoplast region [6,71,72]. Recent studies on the expression of ER-targeted fluorescence proteins have further described the dynamic behavior of the ER in living cells [8,51]. Generally, the complex cortical ER-network structures are found to depend on MFs in plant cells [73,74] but on MTs in animal cells [75,76]. On the other hand, ER-organization in mitotic spindles is associated mainly with MTs rather than MFs in mitotic cells of the gymnosperms, Pinus brutia and Pinus nigra [77], and in tobacco NT-1 cells [51]. Our observation that the tubular ER aligned with mitotic spindles in BA-treated cells (Fig. 7C, 0 min) supports these earlier reports. In addition, we found that BA treatment inhibited ER-accumulation in the phragmoplast, especially in the late expansion phase (Fig. 7C, 48-63 min, D and E), suggesting that MFs are tightly associated with ER recruitment in the phragmoplast. In dividing tobacco mesophyll protoplasts, an unbiased ER redistribution was also found to be dependent on MFs [78].
The ER has been implicated in the control of the cytosolic calcium ion concentrations [79] and in facilitation of membrane exchange between the ER and cell plate during cell plate maturation [6]. Therefore, it is likely that MFs contribute to cell plate expansion via recruitment of ER into the phragmoplast, and this is supported by the observed increase in MF contribution at the late phase. Taken together with the results from endosomal dynamics, these findings thus imply that MFs have multiple functions in the endomembrane systems during cytokinesis. Future studies will need to focus on molecular dissection of the complicated interactions between MFs and endomembranes to gain further insights into MF contribution to plant cytokinesis.

Plant material and cell synchronization
Tobacco BY-2 (Nicotiana tabacum L. cv. Bright Yellow 2) cells and BY-GF11 cells expressing GFP-ABD2 [37] were diluted 95-fold with a modified Linsmaier and Skoog medium at weekly intervals [80]. The cell suspensions were agitated on a rotary shaker at 130 rpm at 27°C in the dark. For cell synchronization, 10 ml of 7-day-old cells were transferred to 95 ml fresh medium and cultured for 24 h with 5 mg liter -1 aphidicolin (Sigma Chemical Co., St. Louis, MO, USA) [81]. The cells were washed with 10 volumes of fresh medium and resuspended in the same medium.

Microscopy
The cells were transferred into φ35 mm Petri dishes with φ14 mm coverslip windows at the bottom (Matsunami, Osaka, Japan). The dishes were placed onto the inverted platform of a fluorescence microscope (IX-70; Olympus) equipped with a CSU10 scanning head (Yokogawa, Tokyo, Japan) and a cooled CCD camera head system (CoolSNAP HQ, PhotoMetrics). For multifocal observations, serial optical sections were taken at 0.5-μm intervals.

Inhibitor
For disruption of MFs, cells were treated with 1 μM bistheonellide A (BA, Wako Pure Chemical Ind., Osaka, Japan) for 60 min before observations. Myosin activity was inhibited by treatment with 20 mM BDM (Sigma Chemical Co., St. Louis, MO, USA).

General Image analysis
Indirect quantification of fluorescence levels, maximum intensity projections and kymograph construction were performed using ImageJ software [46]. Kymographs were constructed with our in-house developed plug-in package for ImageJ software, which is freely available from our website http://hasezawa.ib.k.u-tokyo.ac.jp/zp/Kbi/Image JKbiPlugins.

Time-sequential measurements of cell plate diameter
To measure the diameter of an expanding cell plate, a binary image was obtained by intensity thresholding and used for measurements on ImageJ. A Gaussian filter was applied to the cross-sectional images before binarization. Cell plate diameters were calculated with ImageJ software. Regression analysis of the cell plate diameter was performed using Kaleida-graph software (Synergy Software, Reading, PA).

Assessment of cell plate morphology
To assess the cell plate structures, we first obtained crosssectional images from 70 serial optical sections taken at 0.5 μm intervals of each cell with view orthogonal planes function in MetaMorph image-analysis software (Universal Imaging Co., Downingtown, Panama). Subsequently, we obtained the binary images from the cross-sectional images by intensity thresholding using ImageJ software as shown in Figure 5A. To improve the signal to noise ratio, a Gaussian filter was applied to the cross-sectional images before binarization. Cell plate area and perimeter were calculated from the binary images with ImageJ software. Complexity, a secondary morphological parameter, was calculated by P 2 /4πA, where P is the perimeter and A is the area, according to Tanaka et al. (2007) [47].

Analysis of endosomal movement
To estimate the velocity of endosomal movement, we performed time-sequential observations at 1-sec intervals for 30 sec by spinning-disc confocal laser microscopy. From the time-sequential images, we measured the distance and time traveled by endosomes using the ImageJ plug-in MTrack2, which was developed for multiple particle detection and tracking, and which is freely available from the website http://valelab.ucsf.edu/~nico/IJplugins/ MTrack2.html. Subsequently, endosomal velocity was calculated by dividing the distance traveled by the traveling time.