The actin cytoskeleton of higher plants plays an essential role in plant morphogenesis and in maintaining various cellular activities. In this study we established a tobacco BY-2 cell line, stably transformed with a GFP–fimbrin actin-binding domain (ABD) 2 construct, that allows visualization of actin microfilaments (MFs) in living cells. Using this cell line, designated BY-GF11, we performed time-sequential observations of MF dynamics during cell-cycle progression. Detailed analyses revealed the appearance of a broad MF band in the late G2 phase that separated to form a structure corresponding to the so-called actin-depleted zone (ADZ) in mitosis. In BY-GF11, the MF structure at the cell cortex in mitosis appeared to form two bands rather than the ADZ. Measurements of fluorescent intensities of the cell cortex indicated an MF distribution that resembled two peaks, and we therefore named the structure MF ‘twin peaks’ (MFTP). The cell plate formed exactly within the valley between the MFTP at cytokinesis, and this cell-plate guidance was distorted by disruption of the MFTP by an inhibitor of actin polymerization. These results suggest that the MFTP originates from the broad MF band in the G2 phase and functions as a marker of cytokinesis.
The actin cytoskeleton of higher plants plays essential roles in various aspects of plant morphogenesis, including cell division, pollen-tube growth, trichome development and the formation of stomatal complexes. The cytoskeleton also maintains diverse cellular activities such as cytoplasmic streaming, organellar inheritance and membrane trafficking. In numerous studies, the actin cytoskeleton of different plant cell types has been visualized by fluorescent-phalloidin labelling or immunostaining after fixation. However, because of the difficulties in preserving the integrity of the actin microfilaments (MFs), due to their sensitivity to the chemical fixation treatments, such post-fixation staining methods appear to generate artificial MF rearrangements. An alternative approach to observing MFs, which negates the need for fixation, is the microinjection of fluorescent-labelled phalloidin. Using this technique, the dynamic redistribution of MFs during mitosis and cytokinesis have been clarified in vivo in Haemanthus endosperm cells (Schmit and Lambert, 1990) and in Tradescantia stamen hair cells and stomatal complexes (Cleary, 1995; Cleary et al., 1992; Zhang et al., 1993). However, drawbacks of this technique include the brevity of the observation period due to the diffusion of fluorescence into vacuoles and adjacent cells, and the fact that not all plant cell types are amenable to microinjection. In contrast, recent molecular techniques employing the green fluorescent protein (GFP) have enabled us to visualize MFs in vivo. Direct labelling of actin with GFP has demonstrated MFs in yeast and Dictyostelium (Aizawa et al., 1997; Doyle and Botstein, 1996), although expression of GFP-actin was found to inhibit cellular activities in Dictyostelium cells. An alternative to direct labelling is the labelling of proteins with actin-binding domains (ABDs). GFP fusion to a mouse talin (mTn) actin-binding domain labelled the MFs in tobacco BY-2 cells following particle bombardment, and also those in stably transformed Arabidopsis plants (Kost et al., 1998). Although no time-sequential observations were performed with the above tobacco cells, the fusion of mTn to the fluorescent protein allowed visualization of MFs in pollen tubes and trichomes of Arabidopsis plants (Brembu et al., 2004; Kost et al., 1998; Mathur et al., 1999; Saedler et al., 2004). GFP fusion with the ABD of fimbrin also labelled MFs in tobacco protoplasts and BY-2 cells, Arabidopsis and Medicago cells (Sheahan et al., 2004a,b).
During cell-cycle progression, a dynamic change in the actin cytoskeleton has been demonstrated by fluorescent-phalloidin labelling, after fixation or by microinjection, in various plant cell types such as Allium, tobacco BY-2 and Tradescantia stamen hair cells (Cleary et al., 1992; Hasezawa et al., 1991; Liu and Palevitz, 1992; Mineyuki and Palevitz, 1990). These reports described two main MF structures from the G2 phase to mitosis: one constitutes the actin band that co-localizes with, but is wider than, the microtubule preprophase band (MT PPB, Liu and Palevitz, 1992; Mineyuki, 1999); the other is the so-called actin-depleted zone (ADZ) that appears in late G2 to mitosis (Cleary et al., 1992; Liu and Palevitz, 1992). Although the ADZ has been proposed to determine the position of the cell division plane (Cleary, 1995; Hoshino et al., 2003), the exact relationship between the broad actin band, the ADZ and cell plate has not yet been clarified because of difficulties associated with time-sequential observations in fixed or microinjected cells.
In this study we established a tobacco BY-2 cell line, stably transformed with a GFP–fimbrin ABD2 construct, that allows visualization of MFs in living cells without the need for fixation. Using this cell line, designated BY-GF11, we performed time-sequential observations of MF dynamics during cell-cycle progression. Detailed observations revealed that the broad MF band observed in the late G2 phase separated into two MF bands that were quite similar to the ADZ at mitosis, and that a cell plate formed between these two MF bands during cytokinesis. Our results demonstrate that the two MF bands that appear in mitosis originate from the broad MF band in the late G2 phase, and are involved in determination of the cell plate position.
Visualization of actin microfilaments by stable GFP–fimbrin expression
The second actin-binding domain (ABD2) of the Arabidopsis AtFim1 protein was fused to the C-terminus of GFP and stably expressed in tobacco BY-2 cells to generate the BY-GF11 (BY-2 cells stably expressing GFP–fimbrin line 11) cell line. Confocal observations revealed GFP fluorescence of filamentous structures near the cell cortex and along the cytoplasmic strands at the mid-plane (Figure 1a,b). Such filamentous structures were also observed in another cell line of BY-GF8 transformed with the same construct (see Figure 4). These cortical filamentous structures closely resembled those observed in cells following particle bombardment with the mouse talin F-actin-binding domain (GFP-mTn) (Figure 1c; Kost et al., 1998), although the MF structures in the cytoplasmic strands could not be clearly observed in the GFP-mTn-bombarded cells (Figure 1d). Arabidopsis cell cultures stably expressing GFP–fimbrin ABD2 also revealed similar filamentous structures near the cell cortex and in the cytoplasm (Figure 1e,f). To ensure the GFP-labelled fluorescent structures of the BY-GF11 cells did represent MFs, rhodamine-phalloidin staining was performed after gentle permeabilization of the cells with glycerol. The structures labelled by GFP and rhodamine fluorescence almost overlapped at both the cell cortex and the cell mid-plane (Figure 1g–l). In addition, the GFP fluorescence disappeared following treatment with 100 μm cytochalasin D (CD), an inhibitor of actin polymerization (see Figure 7). These observations clearly indicate that the GFP–fimbrin ABD2 fusion protein labelled MFs in the tobacco BY-GF11 cells.
Cortical and cytoplasmic MF organization in interphases of living tobacco BY-2 cells
As the MF structures were shown to be dynamically changing during cell-cycle progression in fixed cells, we first monitored the changes in MF structures in living BY-GF11 cells. At the G1 phase, transverse and meshwork-like MFs were observed at the cell cortex (Figure 2a), as also found in rhodamin-phalloidin-stained BY-2 cells after fixation (Hasezawa et al., 1991). At the mid-plane, the MFs extended through the cytoplasm from the cell nucleus to the cell periphery (Figure 2b). Notably, MFs were localized at the periphery of the cytoplasmic strands rather than in their middle. A 3-D image constructed by volume rendering showed a basket-like MF structure within a whole cell (Figure 2c). At S phase, when the cell nucleus moved from the cell periphery to the centre, the MFs extended radially from the nuclear surface to the cell periphery (Figure 2e,f). In the G2 phase, the cortical MFs became predominantly transverse-oriented and concentrated at the cell centre so as to form a band-like structure (Figure 2g,i). At the mid-plane, there was an increase in the amount of GFP fluorescence surrounding the cell nucleus and at the mid-region of the cell cortex (Figure 2h). In contrast, the fluorescence along the cytoplasmic strands and at the cortex of both cell ends was weak, and the filaments appeared to be finer than those in the G1 and S phases.
Cortical MFs in the G2 phase separated in mitosis
Subsequent observation of MF structures during mitosis revealed that the cortical MFs that were transversely oriented at the G2 phase formed meshwork structures at metaphase (Figure 2j). These meshwork structures were probably localized on both the plasma and vacuolar membranes (Figure 2k, arrow and arrowhead, respectively). At the centre of the cell, a zone with a weak MF network but still with some MFs localized could be distinguished (Figure 2l, arrowhead). This structure closely resembled the ADZ, although in the BY-GF11 cells there appeared to be two MF bands (Figure 2l, asterisks) rather than the ADZ. At the mid-plane, the MFs stretched throughout the cytoplasm as in the G1 through G2 phases, but the filaments were now very fine (Figure 2k). Only a dim fluorescence was observed in the mitotic apparatus. In late anaphase, two populations of longitudinal MFs appeared temporally between the two sets of chromosomes (Figure 2n), as reported in rhodamine-phalloidin injected Tradescantia cells (Cleary et al., 1992). In telophase, the MFs co-localized with the MT phragmoplast and, according to phragmoplast development, moved to the cell periphery (Figure 2q,r).
The above-mentioned MF dynamics could be observed in a single living cell by time-sequential observations from late G2 through mitosis to cytokinesis. In the late G2 phase, the cortical MF band could be recognized at the centre of the cell (Figure 3a). As the cell cycle progressed, a zone with weak MFs appeared at the centre of the cortical MF band (Figure 3c), then two MF bands became evident at mitosis (Figure 3d–f). After cytokinesis, the cell plate seemed to form at the site of the weak MFs in mitosis (Figure 3i).
As the above time-sequential observations at the mid-plane by confocal laser scanning microscopy (CLSM) suggested a separation of the cortical MF band from the late G2 phase to mitosis, we attempted to observe the dynamics of the cortical MF structures in more detail. The MF band was observed in the late G2 phase, when a phragmosome structure could be clearly distinguished by a differential interference contrast (DIC) image obtained from light microscopy (Figure 4a,e). As the cell cycle progressed to mitosis, the MF band began to separate at the cell cortex, and a zone with a weak cortical MF network appeared at the centre of the MF band (Figure 4f–h). Measurements of fluorescent intensity at the cell cortex by fluorescent microscopy showed the highest intensity at the centre of the cell in the late G2 phase (Figure 4i). With cell-cycle progression, the fluorescent intensity at the cell centre decreased (Figure 4j) and resulted in the appearance of two peaks together with an increase in the distance between the MF band(s) (Figure 4k,l). In accordance with the reduced fluorescent intensity at the cell centre, the phragmosome structure became obscure (Figure 4d). Based on the distribution of fluorescent intensities of the cortical MF structures in mitosis, we named the structure composed of the two MF bands as the MF ‘twin peaks’ (MFTP). This MFTP structure was confirmed in another cell line of BY-GF8 and in BY-2 cells incubated with rhodamine-phalloidin after permeabilization (Figure 4m–r).
A time-sequential observation at the cell surface also revealed the separation of the MF band (Figure 5). In the late G2 phase, MFs that were transversely oriented (Figure 5b) started to develop a meshwork structure and subsequently formed two MF bands with a weak MF network zone in mitosis (Figure 5h).
Relationship of MFTP and cell plate guidance
Continuation of the time-sequential observations to the G1 phase demonstrated that the cell plate was formed 120 min from the G2 phase (Figure 6a). The cell plate appeared to develop at the position previously occupied by the zone with a weak MF network in metaphase (Figure 6c). Measurement of the distances between the two peaks of the MFTP at mitosis, and also of that between one peak and the cell plate, indicated that the latter was almost half the distance of the former (Figure 6e), and suggested that the cell plate was formed in the valley of the MFTP.
To investigate the functional relevance of the MFTP to the cell plate guidance, we examined an effect of cytochalasin D (CD), an inhibitor of actin polymerization. When we applied CD (100 μm) 6 h after aphidicolin removal, the MF structures disappeared in 2 h (Figure 7a). Although cell-cycle progression was not affected by the CD treatment (Figure 7b), abnormal cell plates were observed 11 h after aphidicolin removal by aniline blue staining (Figure 7e). In the non-treated control cells the cell plates were smooth and sharp, while in the CD treated cells some were not vertical to the cell cortex and some were broad and twisted. Removal of the CD at 8 h restored the MF structures, but the cell plate was still twisted (Figure 7a,e). These ‘distorted’ cell plates were observed for approximately 30% of the cells treated with CD from 6 to 8 h, in which the cells with MFTP were mostly observed (Figure 7c,d). In contrast, in cells treated with CD from 9 to 11 h, in which phragmoplasts were mostly observed, the percentage of distorted cell plates decreased significantly (Figure 7c,d).
Establishment of tobacco BY-GF cell line
In this study we established a tobacco BY-GF11 cell line expressing a GFP–fimbrin ABD2 fusion protein through which MFs could be visualized, and the MF dynamics during cell-cycle progression could be observed. In previous studies, an ABD of mouse talin (mTn) was most commonly used to visualize MFs in living cells (Kost et al., 1998). Microfilaments were observed during trichome development of a stable GFP-mTn-transformed Arabidopsis line (Brembu et al., 2004; Saedler et al., 2004), but some toxic effects of mTn were also suggested (Ketelaar et al., 2004). Our transient expression analyses using particle-bombarded GFP-mTn revealed the presence of cortical MFs, but no clear cytoplasmic MF structures could be observed (Figure 1c,d). The development of a stable GFP-mTn-transformed tobacco BY-2 cell line was reported recently, although no time-sequential observations have yet been demonstrated (Hoffmann and Nebenführ, 2004). On the other hand, our BY-GF11 cells that stably express the GFP–fimbrin ABD2 fusion protein clearly show the MF structures, including those within cytoplasmic strands, during cell-cycle progression; these MF structures were confirmed in another cell line of BY-GF8 (data not shown). The BY-GF11 cells grow as well as non-transformed tobacco BY-2 cells, and show a cell synchrony of approximately 50% of the mitotic index after aphidicolin removal (Figure 7b). The period from late G2 through mitosis to the G1 phase in a single living cell was similar to that observed in GFP-tubulin- and GFP-AtVam3-transformed BY-2 cells (Kumagai et al., 2001; Kutsuna and Hasezawa, 2002).
Fimbrin is an actin-binding protein, and the first plant fimbrin-like gene, AtFIM1, was identified in Arabidopsis (McCurdy and Kim, 1998). Whereas GFP fusions with the full-length AtFim1 protein, or with the amino terminus of the ABD1, did not show clear MF structures (Sheahan et al., 2004b), fusions with the fimbrin ABD2 enabled visualization of MFs in tobacco protoplasts and BY-2 cells, Arabidopsis, and also Medicago cells (Sheahan et al., 2004a,b). Our BY-GF11 cell line, expressing a GFP–fimbrin ABD2 fusion protein, also successfully allowed visualization of MFs. Moderate binding activity of the fimbrin ABD2 to MFs was suggested to provide a non-invasive level of MF labelling (Sheahan et al., 2004b). Our observation that the GFP fluorescence of BY-GF11 cells could be maintained only after treatment with a low concentration of glycerol to gently permeabilize the plasma membrane (Figure 1g,h) supports the affinity of the fimbrin ABD2 for MFs.
Observation of actin MFs in BY-GF11 cells
The MF structures observed in BY-GF11 cells were essentially similar to those visualized in tobacco BY-2 cells with fluorescent phalloidin after fixation (Hasezawa et al., 1991; Hoshino et al., 2003). Rhodamine-phalloidin labelling is thought to bundle the MFs: thick MFs were often observed, especially without treatment with m-maleimidobenzoyl N-hydroxysuccinimide ester (MBS), a potent protein bridging reagent (Hasezawa et al., 1989; Sonobe and Shibaoka, 1989). In contrast, thick MFs were observed in the BY-GF11 cells from G1 to S phase, whereas very fine MFs were identified from G2 to mitosis. Although the precise role of these thick and fine MFs has yet to be clarified, the fine MFs could be visualized with GFP–fimbrin ABD2 without rhodamine-phalloidin labelling.
In rhodamine-phalloidin staining after fixation, some winding filaments or radially expanded filaments could be observed to extend from the cell nucleus to the cell periphery (Hasezawa et al., 1991). In contrast, in living BY-GF11 cells, the MFs localized along the cytoplasmic strands (Figures 1b and 2b) and appeared to interact with the tonoplast. This difference may be due to disruption of the membrane structures by the fixation process. Microfilaments visualized by microinjection showed a cytoplasmic localization similar to our BY-GF11 cells, and suggested an interaction of the MFs with the tonoplast (Cleary et al., 1992). A similar conclusion was reached by observations of a tobacco BY-GV7 cell line, expressing a GFP-AtVam3 fusion protein that allows visualization of the tonoplast in living cells, following rhodamine-phalloidin staining after saponin treatment (Kutsuna et al., 2003). This interaction therefore suggests that either the MFs maintain their structures by interacting with the tonoplast, or the MFs play a role in morphogenesis of the large vacuole.
Formation of MFTP and its role in the cell plate guidance
The BY-GF11 cell line allowed us to observe the time-sequential dynamics of MF structures, and revealed a separation of the cortical MF band from the late G2 phase (Figures 3–5). Changes in distribution of the relative fluorescence intensity at the cell cortex showed this separation clearly (Figure 4i–l), and allowed us to identify the resulting MF structure which we termed the MFTP. The separation formed two MF bands, together with a region of weak GFP fluorescence that closely resembled the so-called ADZ. Observations of the BY-GF cells showed the cortical MFs to be distributed only in the central region, and apparently to form two bands. Although the relationship between the cortical MF bands and ADZ has not been completely addressed, Mineyuki (1999) suggested the appearance of an ADZ within the cortical MF band. Our observations of a living cell clearly demonstrated that the cortical MF band was the origin of the MFTP, and that the valley between the peaks closely resembled the ADZ.
The MFs in the cortical MF band at the G2 phase were mainly transverse-oriented, whereas those after separation showed meshwork-like structures (Figure 5). These MF structures therefore appear to have been formed not by a simple separation and movement of the transverse cortical MFs, but rather by some rearrangement of the MF networks.
The ADZ has been suggested to function as a marker of cytokinesis after PPB disruption (Cleary, 1995; Hoshino et al., 2003), and our time-sequential observations further suggest that the site of the cell plate corresponds to the valley of the MFTP (Figure 6c). Measurements of fluorescent intensities of cortical MFs strongly support the above observation. The increased number of cells with a distorted cell plate by disruption of the MFTP suggested its role in proper cell plate guidance (Figure 7d). Under conditions in which two PPBs were formed in the G2 phase, the cell plate was mostly formed so as to connect the two PPBs (Hasezawa et al., 1994). In these cells, two ADZs were observed at the sites of the PPBs and disruption of the MF structure released the connection (Yoneda et al., 2005). This observation will support the role of the MFTP as a marker of cell plate positioning in cytokinesis.
Interestingly, disruption of the actin phragmoplast had little effect on cell plate guidance (Figure 7c,d). The formation and orientation of the cell plate appear to be independent processes. Our results also suggest that, in telophase, the MF structures were not necessary for cell plate guidance. The positional information provided by MFTP will be left as a ‘memory’; however, substantial evidence has not yet been clarified.
In conclusion, we have established a tobacco BY-GF11 cell line that expresses GFP–fimbrin ABD2 and allows visualization of the MF structures. Time-sequential observations demonstrated the appearance of MFTP, which formed a valley closely resembling the ADZ, and which originated from the cortical MF band in the late G2 phase. Subsequent cell plate formation at exactly the site occupied by the MFTP valley at mitosis further strengthens the proposed role of MFTP in determining the division site. We are certain that the use of the BY-GF11 line in future studies will provide further insight not only into the MF structures during cell-cycle progression, but also into other cellular events such as the role of MFs in vacuolar morphogenesis.
Tobacco BY-2 (Nicotiana tabacum L. cv. Bright Yellow 2) suspensions were diluted 95-fold with a modified Linsmaier and Skoog medium supplemented with 2,4-D (LSD) at weekly intervals, as described by Nagata et al. (1992). The cell suspensions were agitated on a rotary shaker at 130 rpm at 27°C in the dark. Cell synchronization was performed as described by Nagata et al. (1992). In brief, 10 ml 7-day-old cells were transferred to 95 ml fresh medium and cultured for 24 h with 5 mg l−1 aphidicolin (Sigma, St Louis, MO, USA). The cells were washed with 10 vol fresh medium and then resuspended in the same medium. After aphidicolin release, cell-cycle progression was monitored by counting the percentage of cells in mitosis (mitotic index), which was determined with a fluorescence microscope after staining the nuclei with 20 μg l−14′,6-diamidino-2-phenylindole (DAPI).
Construction of GFP–fimbrin ABD2
The region encoding the second actin-binding domain (ABD2; amino acids 325–687) of AtFim1 (McCurdy and Kim, 1998) was PCR amplified from an Arabidopsis thaliana cDNA library. The amplified fragment replaced the mTn of the CaMV 35S-sGFP (S65T)-mTn vector, kindly provided by Dr T. Kagawa of Tsukuba University, resulting in an in-frame fusion of N-terminus GFP and C-terminus fimbrin ABD2. The 35S-sGFP (S65T)-mTn vector was modified from the 35S-sGFP (S65T) vector, obtained from Dr Y. Niwa of Shizuoka Prefecture University, so as to contain 6 × Gly-Ala repeats between GFP and the mTn. The resulting 35S-sGFP (S65T)-fimbrin ABD2 region was then subcloned into the pCAMBIA1300 binary vector (CAMBIA, Canberra, Australia) for transformation.
Transformation and establishment of tobacco BY-2 and Arabidopsis cell lines stably expressing the GFP–fimbrin fusion protein
The GFP–fimbrin construct was transformed into Agrobacterium tumefaciens strain, LBA4404. A 4-ml aliquot of 3-day-old BY-2 cells was incubated with 100 μl overnight culture of the transformed A. tumefaciens as described by An (1985). After 2 days’ incubation at 27°C, cells were washed four times in 5 ml LSD medium, then plated onto solid LSD medium containing 500 mg l−1 carbenicillin and 15 mg l−1 hygromycin. Calluses, which appeared after 20 days, were transferred onto new plates and cultured independently until they reached approximately 1 cm in diameter, when they were transferred to 20 ml liquid LSD medium and agitated on a rotary shaker at 130 rpm at 27°C in the dark. After 1 month a cell line suitable for observing MFs was selected by examination of GFP-fluorescence by fluorescent microscopy, and the cell line obtained was designated BY-GF11.
Transformation of cell suspensions of Arabidopsis Col-0 was performed essentially as described by Oda et al. (2005) with the same GFP–fimbrin ABD2 construct.
A cell suspension of 3-day-old BY-2 cells was filtrated onto filter paper, and the cells bombarded with gold particles (1.0 μm) coated with the appropriate vector constructs using a particle-delivery system (PDS-1000/He, Bio-Rad, Hercules, CA, USA) according to the manufacturer's recommendations. Filtrated BY-2 cells were placed at a distance of 6 cm under the stopping screen and bombarded in a vacuum of 28 inches Hg at a helium pressure of 1100 psi. Following bombardment, cells were diluted in LSD medium and kept in the dark at 27°C for 6–12 h before observation.
For simultaneous observations of rhodamine-phalloidin and GFP–fimbrin, the BY-GF11 cells were suspended in a solution containing 50 mm PIPES (pH 6.8), 1 mm MgSO4, 5 mm EGTA, 1.5% glycerol, 150 mm mannitol and 0.07 μm rhodamine-phalloidin (Molecular Probes, Eugene, OR, USA) and incubated for 30 min at room temperature.
For time-sequential observations, synchronized BY-GF11 cells were transferred into 35-mm-diameter Petri dishes with 14-mm-diameter coverslip windows at the bottom (Matsunami Glass Ind. Ltd, Osaka, Japan). The dishes were placed onto the inverted platform of a fluorescence microscope equipped with a confocal laser scanning head and control systems (CLSM GB-200; Olympus, Tokyo, Japan) or a cooled CCD camera head system (Cool-SNAP HQ; PhotoMetrics, Huntington Beach, Canada). Three-dimensional images were reconstructed from the continuous optical sections within a 0.5- μm step size using metamorph software (Universal Imaging, Downingtown, PA, USA). The images were processed digitally using photoshop software (Adobe, San Jose, CA, USA).
On request, all novel materials described in this publication will be made available in a timely manner for non-commercial research purposes.
We are grateful to Dr T. Kagawa (Tsukuba University) for the kind gift of the 35S-sGFP(S65T)-mTn vector and to Dr Y. Sato (National Institute for Basic Biology) for technical advice. We thank Mr H. Hoshino (University of Tokyo) for preliminary experiments and Dr F. Kumagai-Sano (Gunma University) for a critical reading of the manuscript. This study was supported in part by a Grant-in-Aid for Scientific Research on Priority Areas (Grant No. 15031209) to S.H. from the Ministry of Education, Culture, Sports, Science and Technology, Japan.