- American Society of Plant Biologists
Abstract
In a previous study on Arabidopsis thaliana suspension cells transiently infected with the microtubule end binding protein AtEB1a–green fluorescent protein (GFP), we reported that interphase microtubules grow from multiple sites dispersed over the cortex, with plus ends forming the characteristic comet-like pattern. In this study, AtEB1a-GFP was used to study the transitions of microtubule arrays throughout the division cycle of cells lacking a defined centrosome. During division, the dispersed origin of microtubules was replaced by a more focused pattern with the plus end comets growing away from sites associated with the nuclear periphery. The mitotic spindle then evolved in two quite distinct ways depending on the presence or absence of the preprophase band (PPB): the cells displaying outside-in as well as inside-out mitotic pathways. In those cells possessing a PPB, the fusion protein labeled material at the nuclear periphery that segregated into two polar caps, perpendicular to the PPB, before nuclear envelope breakdown (NEBD). These polar caps then marked the spindle poles upon NEBD. However, in the population of cells without PPBs, there was no prepolarization of material at the nuclear envelope before NEBD, and the bipolar spindle only emerged clearly after NEBD. Such cells had variable spindle orientations and enhanced phragmoplast mobility, suggesting that the PPB is involved in a polarization event that promotes early spindle pole morphogenesis and subsequent positional stability during division. Astral-like microtubules are not usually prominent in plant cells, but they are clearly seen in these Arabidopsis cells, and we hypothesize that they may be involved in orienting the division plane, particularly where the plane is not determined before division.
INTRODUCTION
In higher plant cells, the origins of the different microtubule arrays are not always clear because these cells do not contain an obvious microtubule organizing center, like the animal centrosome. Higher plant cells form four different microtubule arrays that succeed each other during the cell cycle: the interphase cortical array, the preprophase band (PPB), the mitotic spindle, and the cytokinetic phragmoplast. During interphase, microtubules are diffusely distributed at the cell cortex, where they associate with the plasma membrane and form parallel groups. At the plasma membrane, cortical microtubules are motile, exhibiting a hybrid form of dynamic instability and treadmilling (Shaw et al., 2003), and we recently showed that microtubules grow and shrink from multiple foci that can be tagged with Arabidopsis thaliana end binding protein 1a–green fluorescent protein (AtEB1a-GFP) (Chan et al., 2003).
During preprophase, the PPB develops from the selective stabilization and destabilization of cortical microtubules (Dhonukshe and Gadella, 2003), which gives rise to a band of cortical microtubules that encircles the nucleus. The directionality of EB1 comets shows that the PPB, like the cortical array, is composed of microtubules of mixed polarity (Chan et al., 2003). Despite appearing before mitosis, the PPB has a connection with cytokinesis in that it marks the plane of cell division (Mineyuki, 1999). In addition to the cortical band that somehow marks the division plane, the microtubules radiating from the preprophase nuclear surface have also been suggested to play roles in premitotic nuclear positioning, spindle orientation, and spindle pole morphogenesis (Lloyd, 1991; Mineyuki, 1999; Granger and Cyr, 2000). The fact that microtubules radiate from the nuclear envelope during preprophase (also in the early postcytokinetic period) has led to the nuclear periphery being one of the accepted sites for microtubule nucleation in plants (Erhardt et al., 2002; Schmit, 2002).
The next transition concerns the formation of the spindle. At prophase, a cage of microtubules develops around the nucleus, forming the prophase spindle which, by metaphase, consists of two half sets of microtubules with opposite polarity. The plus ends of the microtubules for each half set lie in the equatorial plane (Euteneuer et al., 1982). The spindle poles are not as tightly focused as in animal cells for there may be multiple minipoles that frequently give rise to broad poles. At anaphase, after the chromosomes have separated, the microtubules of the spindle midzone give way to a column comprised of two sets of interdigitating microtubules; in the conventional cytokinesis of somatic cells, these remnants of the central spindle are considered to initiate the cytokinetic apparatus (Staehelin and Hepler, 1996; Otegui and Staehelin, 2000). This, the phragmoplast, then grows out centrifugally, opening up as a ring that deposits the cell plate in its center. Microtubule hook decoration studies have shown that phragmoplast microtubules share the same polarity as those of the spindle, with the plus ends overlapping in the equatorial plane (Euteneuer and McIntosh, 1980). Drug and molecular biological studies indicate that phragmoplast expansion is dependent upon continuous microtubule depolymerization and repolymerization, during which microtubule disassembly is tightly coupled to cell plate expansion (Yasuhara et al., 1993; Nishihama and Machida, 2001).
Throughout this cyclic formation of different microtubule arrays, the absence of a centriole-containing centrosome as an unambiguous marker of microtubule nucleation has led to uncertainty about the origin of, and continuity between, the various microtubule arrays. For instance, the absence of centrosomes from the plant cell's spindle poles has been taken as support for the idea that spindles form by an inside-out mechanism in which microtubules first associate with the chromosomes before being sorted and drawn by microtubule motors into an antiparallel array in which the minus ends only latterly become organized into poles (reviewed in Compton, 2000). The origins and sites of microtubule nucleation during phragmoplast expansion are also unclear. In somatic cells, the phragmoplast appears to evolve out of spindle remnants (Zhang et al., 1990; Otegui and Staehelin, 2000), whereas in endosperm, the phragmoplast microtubules are nucleated from the surfaces of the daughter nuclei with the cell plate being deposited where microtubules overlap at the periphery of the various nuclear-cytoplasmic domains (Vaughn and Harper, 1998; Brown and Lemmon, 2001). However, in somatic cells, once the phragmoplast grows out beyond the internuclear corridor, it is not apparent where new microtubules originate, and the idea that microtubule plus ends overlap at the midline is not universally agreed upon: the possibility has been raised that nucleating material becomes deposited in the midline, resulting in a reversal of microtubule polarity (Schmit, 2002). Markers for the plus or the minus ends of microtubules should help us to understand the origin of the arrays.
Although higher land plants do not have centrioles as part of their life cycle, it is possible that they possess some of the antigens that constitute the amorphous pericentriolar material from which spindle and cytoplasmic microtubules emerge. γ-Tubulin is an important constituent of animal centrosomes, forming ring complexes that act as templates for new microtubule growth (Zheng et al., 1995). However, in plants, γ-tubulin is found in variously sized complexes (Drykova et al., 2003): it decorates the entire length of microtubules, not just the minus ends, and has consequently not proved a good marker for nucleation sites (Joshi and Palevitz, 1996).
An alternative marker for microtubule ends is the end binding protein EB1, a member of a conserved family of microtubule-associated proteins that in other eukaryotic cells is known to bind the plus ends of microtubules. At the plus end of microtubules, EB1 regulates microtubule dynamics by acting as an anticatastrophe factor (Rogers et al., 2002; Tirnauer et al., 2002) and promotes the end-on-attachment of microtubules to different subcellular targets (Korinek et al., 2000). However, EB1 has also been shown to accumulate at the minus ends of microtubules (Bu and Su, 2001; Morrison and Askham, 2001; Askham et al., 2002; Rehberg and Gräf, 2002; Rogers et al., 2002; Tirnauer et al., 2002, 2004; Louie et al., 2004; Piehl et al., 2004) in organisms as diverse as fungi (Straube et al., 2003) to vertebrates (Piehl et al., 2004). A plausible explanation is that the minus end EB1 acts as a repository for the material that grows out with the plus end (Rehberg and Gräf, 2002).
Recently, we reported on the distribution of an Arabidopsis homolog of EB1 (AtEB1a) transiently expressed as a C-terminal GFP fusion in Arabidopsis suspension cells (Chan et al., 2003). We demonstrated that AtEB1a-GFP marks the plus ends of microtubules with the familiar comet-like pattern (see also, Mathur et al., 2003; van Damme et al., 2004). Using time-lapse microscopy, we also showed that AtEB1a-GFP marked foci at the minus ends from which microtubules (sometimes more than one) were observed to grow and shrink back (Chan et al., 2003).
Now, using AtEB1a-GFP as a directional marker, we report on the origin and continuity of the microtubule structures involved in division, showing the existence of alternative pathways for spindle pole and phragmoplast expansion in Arabidopsis suspension cells.
RESULTS
Time-lapse microscopy was used to follow the direction of plus end comet growth and the general staining pattern of AtEB1a-GFP around the cell cycle. A brief overview of the changing patterns of fluorescence is presented before focusing on phases of particular interest.
As previously shown (Chan et al., 2003), during interphase, plus end comets arose at multiple dispersed sites around the cortex, but with the onset of division, the nuclear periphery became a prominent site for microtubule outgrowth. Plus end comets were observed to grow out from the perinuclear region toward the cell cortex (see Supplemental Movie 1 online) to form a radial array, in contrast with the interphase array in which the microtubule ends are dispersed over the cortex. Prophase cells were also readily distinguishable from interphase cells by accumulations of AtEB1a-GFP around the periphery of their nuclear envelopes (Figures 1A and 1B ).
Subcellular Distribution of AtEB1a-GFP in Dividing Suspension Cells of Arabidopsis.
Distribution of AtEB1a-GFP in living suspension cells of Arabidopsis ([A] to [F]) at prophase/preprophase ([A] and [B]), metaphase (C), anaphase (D), and cytokinesis ([E] and [F]). Arrowheads indicate accumulations of AtEB1-GFP at spindle poles ([B] to [D]). Arrows in (E) indicate accumulations around the nucleus. Arrowheads in (F) indicate accumulated label in the midline of the phragmoplast. Bar = 7 μm.
Metaphase spindles were characterized by general labeling of the microtubules with heavy concentrations of AtEB1a-GFP at their spindle poles (Figures 1C and 1D). Double labeling with antitubulin confirmed that the GFP signal was concentrated at the spindle poles (Figures 2A to 2E ), forming annular structures in favorably presented spindles (Figure 2E). Time-lapse movies showed that the dynamic ends of spindle microtubules were distal to these accumulations of AtEB1a-GFP at spindle poles (data not shown).
Triple Staining of Dividing Cells Transformed with AtEB1a-GFP.
Transformed Arabidopsis suspension cells were fixed and stained with anti-GFP antibodies (green), antitubulin antibodies (red), and 4′,6-diamidino-2-phenylindole (DAPI) (blue).
(A) to (E) Metaphase/anaphase cells expressing AtEB1a-GFP:DAPI (A), antitubulin (B), anti-GFP (C), and merge (D). A magnified image of the top, right-hand spindle pole in (C) is shown in (E). The arrow indicates the presence of an annular structure found at the pole.
(F) to (Q) Cells expressing AtEB1a-GFP during cytokinesis: DAPI ([F], [J], and [N]), antitubulin ([G], [K], and [O]), anti-GFP ([H], [L], and [P]), and merge ([I], [M], and [Q]).
Bar = 14 μm in (A) to (D), 5 μm in (E), 13 μm in (F) to (I), and 11 μm in (J) to (M) and (N) to (Q).
During telophase and cytokinesis, AtEB1a-GFP accumulated in two main locations: around the daughter nuclei and in the midzone of the phragmoplast (Figures 1E and 1F). The relationship of these locations to microtubules was confirmed by double labeling fixed cells with antitubulin antibodies (Figures 2F to 2Q). Time-lapse movies established that plus end comets grow away from the opposing faces of the daughter nuclei during telophase (see Supplemental Movie 2 online) and cytokinesis (see Supplemental Movie 3 online). This places the plus ends of microtubules at the midline of the phragmoplast, distal to the forming nuclear surfaces, in agreement with the hook decoration studies of Euteneuer and McIntosh (1980). In favorably presented cells, the comets appeared to originate from foci of AtEB1a-GFP at the nuclear periphery (Figure 3 , focus 1)—often multiple comets emerged from a single focus (see foci 2 and 5). These foci were also highly dynamic, capable of circumferential and radial movement at the nuclear periphery, concomitant with the microtubule polymerization events (foci 2 and 3). In some cases, the foci exhibited a reduced intensity of fluorescence (focus number 5) or disappeared (focus number 1) after the emergence of the comets.
Plus End Comets Emerge from Foci of AtEB1a-GFP at the Periphery of the Nuclear Envelope during Cytokinesis.
AtEB1a-GFP marks foci at the nuclear envelope (arrowheads 1 to 5) from which plus end comets of AtEB1a-GFP (small arrowheads) can be seen to grow toward the cell cortex or phragmoplast. Focus 1 (10 s; [A]) disappears (frame 110 s; [C]) after the emergence of a plus end comet in frame 30 s (B)—microtubule growth then ceases from this spot. Foci 2 and 3 display mobility at the circumference of the nucleus concomitant with microtubule nucleation events—two plus end comets can be seen to emerge from focus 2 and split at different angles in frame 110 s (C). Focus 4 appears in frame 110 s (C) and demonstrates microtubule growth toward the phragmoplast. Focus 5 (frame 215 s; [E]) demonstrates that the brightness of foci decreases after comet growth; similar to focus 2, two comets emerge from focus 5 and grow at divergent angles (frame 235 s; [F]). Note that AtEB1a-GFP is absent from the lower face of the daughter nucleus where no microtubule growth can be detected.
T, time (s); arrowheads, minus end foci of AtEB1a-GFP; small arrowheads, plus end comets; asterisks, outer edges of the phragmoplast; N, upper nucleus of daughter pair. Bar = 10 μm.
Time-lapse movies of anaphase and cytokinesis also showed that the plus end comets contribute to the midline location of AtEB1a-GFP at the phragmoplast. Supplemental Movie 2 online demonstrates that this labeling arose from the plus end comets of microtubules that grew from the opposing faces of the daughter nuclei, meeting at the midline. In actively expanding cells, this midline labeling was limited to the leading edge of the phragmoplast (Figures 4A to 4F ), resulting in a ring-shaped labeling pattern during symmetrical growth of the phragmoplast (Figure 4F); however, less regular patterns were formed during polarized forms of phragmoplast growth (e.g., Figure 1F). No evidence was found for the contrary outgrowth of microtubules from the phragmoplast midline toward the daughter nuclei. These data therefore indicate that the plus ends of growing microtubules are orientated toward the midzone of the phragmoplast.
AtEB1a-GFP Accumulates at the Midline of the Phragmoplast.
Time-lapse confocal laser scanning microscopy images of the anaphase–telophase transition in a single cell show that AtEB1-GFP is present at the midzone of the phragmoplast throughout its morphogenesis ([A] to [F]). Selected stacks highlighting the nucleus (N) are shown in (E). Selected stacks from (D) showing the outer ring staining of the growing phragmoplast, which has rotated, are shown in (F). Arrows indicate nucleus-associated microtubules that interact with the cell cortex. T, time (min). Bar = 6.4 μm in (A) to (F).
Different Mechanisms of Plant Mitosis
By observing microtubule arrays in dividing Arabidopsis cells, we were able to identify a subpopulation of cells that divided without a PPB. Despite the absence of the predictive band of microtubules, the accumulation of label around the nuclear envelope (e.g., Figure 1A) marked these cells as premitotic and allowed them to be followed as they entered mitosis.
Time-lapse studies revealed that there were two pathways for spindle pole development depending on whether a PPB was present or not. In this cell line, ∼70% of cells in prophase did not have a PPB even after the images were contrast stretched using an inverted grayscale image to visualize the cortex more clearly (Figure 5A , T0). All images presented in subsequent parts of the study were taken from cells that were tracked successfully through division (prophase cells representing 0.17% of transformed cells).
AtEB1a-GFP during Spindle Morphogenesis: Alternative Routes to Pole Formation, with and without a PPB.
Arabidopsis suspension cells undergoing spindle morphogenesis in either the absence ([A] to [D]) or presence of a PPB ([E] to [H]). In the absence of a PPB, spindle poles develop during prometaphase after NEBD and redistribution of AtEB1a-GFP around the condensed chromosomes ([C] and [D]). Accumulations of AtEB1a-GFP at spindle poles then become evident at metaphase/anaphase ([D], arrowheads). In the absence of a PPB, the bipolar axis of the spindle is variable, in this case forming perpendicular to the cell's long axis ([A]; the inverted image [T0] highlights the cell's outline). By contrast, when cells display a PPB ([E] to [H]), the bipolar axis of the spindle forms at prophase (G) before NEBD and the mixing of AtEB1a-GFP with the condensed chromosomes. During this process, AtEB1-GFP labels polar caps (G) that mark the poles of the prometaphase spindle after NEBD (H). In cells with a PPB, the bipolar axis of the spindle always forms parallel to the cell's long axis. The arrowheads ([E] to [H]) indicate cortical microtubules of the PPB. Post NEBD indicates that the nuclear envelope has broken down. T, time (min). Bar = 6.1 μm in (A) (9.6 μm in the inverted grayscale image), 7 μm in (B) to (D), and 7.9 μm in (E) to (H).
In cells without a PPB, the premitotic nucleus was associated with radial microtubule arrays. An important characteristic of mitosis in such cells was that spindle bipolarity always developed during prometaphase, after nuclear envelope breakdown (NEBD). This is exemplified in Figures 5A to 5D, where it can be seen that AtEB1a-GFP became diffusely distributed in the vicinity of the condensed chromosomes upon NEBD (Figure 5C). Spindle bipolarity was then expressed during prometaphase, with AtEB1a-GFP becoming concentrated at the spindle poles by metaphase/anaphase (Figure 5D). These data therefore show that cells proceeding into cell division without a PPB did not exhibit distinct, bipolar staining patterns of AtEB1-GFP until the bipolar axis of the spindle had been established after NEBD.
By contrast, in those cells that did display a PPB, spindle bipolarity was established much earlier, at preprophase/prophase (Figures 5E to 5H). In these cases, spindle poles developed from sites at opposite sides of the premitotic nucleus so that the interpolar axis was perpendicular to the PPB (Figure 5G). This polar arrangement of AtEB1-GFP at prophase differed significantly from the nonpolarized form seen in cells without a PPB (cf. Figures 1A with 1B). The fact that polar caps of AtEB1a-GFP coexisted with the PPB suggested that the nuclear envelope was still intact during spindle morphogenesis. This was confirmed in Arabidopsis suspension cells expressing HDEL–yellow fluorescent protein (YFP) and immunostained with antitubulin antibodies (see Supplemental Figure 1 online). This showed that the nuclear envelope was intact when the PPB was present, indicating that the development of spindle bipolarity is flexible, in that it can occur both before and after NEBD.
In addition to differences in spindle pole development, we also found that the orientations of spindles were different in the two subpopulations of cells. In all of the seven cells with a PPB that were tracked through division, the spindle formed in a predictable orientation, perpendicular to the long axis of the cell. This differed significantly to mitosis in the absence of a PPB, in which the premitotic nucleus was asymmetrically disposed and the equatorial plane of the spindle was variable. Of the 11 prophase cells observed by time lapse to undergo spindle formation in the absence of a PPB, eight spindles formed with equatorial planes parallel to the cell's long axis (as shown in Figures 5A to 5D) and two formed with equatorial planes perpendicular to the cell's long axis. The orientation of the remaining spindle was unknown because the cell was spherical in shape. It would therefore appear that in the absence of a PPB, the orientation of the spindle is indeterminate.
Variable Forms of Phragmoplast Expansion
Although two forms of early mitosis were seen in Arabidopsis suspension cells, the genesis of the phragmoplast proceeded by a single route—always occurring from the central spindle region. During the anaphase–telophase transition, the pronounced labeling at the poles declined, and AtEB1a-GFP relocated diffusely to the region of the central spindle and then became more focused again around the reformed nuclei (Figure 6 ).
AtEB1a-GFP during Phragmoplast Morphogenesis.
Time-lapse confocal laser scanning microscopy images of AtEB1a-GFP during the transition between metaphase and telophase in Arabidopsis suspension cells. The phragmoplast develops from the center of the spindle. Arrowheads (T3.12 to T4.00) indicate the presence of an annular structure at the spindle pole. AtEB1a-GFP then concentrates in the vicinity of the daughter nuclei after their reconstitution in telophase. The arrow shows the development of a cytoplasmic strand emanating from the nucleus and then incorporating into the growing face of the phragmoplast. The arrowheads (T14.22 to T15.91) indicate the formation of a new site in the outer margin of the phragmoplast. T, time (min). Bar = 8 μm.
As demonstrated in the movies of telophase cells (see Supplemental Movies 2 and 3 online), we observed microtubule outgrowth to occur from the nuclear periphery during the early stages of phragmoplast morphogenesis. During later stages, when the phragmoplast was expanding, we observed (Figure 6) that nucleus-associated strands of cytoplasm continued to contribute to the growing edge of the cytokinetic apparatus (T10.70 to T12.88). However, in some cells, the nucleus becomes less tightly associated with the phragmoplast and its growing front, with the cytokinetic apparatus appearing to break away during subsequent stages. This can be seen by comparing the relative positions of the lower nucleus in panel T10.70 with T12.83 of Figure 6. In this case, a GFP-labeled locus forms independently of the nuclear surface, with filaments radiating from it toward the midzone of the leading edge of the phragmoplast (Figure 6, panels T14.22 to T15.91); this could provide a potential means of expanding the phragmoplast as it grows beyond the corridor between the daughter nuclei.
The process of centrifugal outgrowth also showed some variability in its behavior. We were able to follow by time lapse both symmetrical (11 cells, e.g., see Figures 3A to 3F) and polarized (five cells, where the phragmoplast contacts one mother wall first, see Cutler and Ehrhardt, 2002) forms of expansion in Arabidopsis suspension cells. The daughter nuclei of these Arabidopsis suspension cells often remained stationary during cytokinesis (e.g., see Figure 6). By contrast, we encountered cases where the nuclei moved. Figure 7 shows a different kind of polarized cytokinesis in which phragmoplast expansion began symmetrically (T0 min) but switched to polarized growth after contact of the cell plate with one side of the cell cortex (T9.46 min). Instead of remaining static, the daughter nuclei then migrated toward the opposite side of the cell in the wake of the outgrowing phragmoplast (T24.0 to T59.21 min). Microtubular fluorescence was then diminished at the side of the phragmoplast that had made first contact with the cell wall and became biased toward the leading edge of the phragmoplast, giving rise to a C-shaped array of microtubules (T24.0 to T59.21 min; note the loss of microtubules from the left-hand sides of both nuclei).
Polarized Cytokinesis.
Time-lapse confocal laser scanning microscopy images of a novel form of polarized cytokinesis. Selected stacks show the mobility of the daughter nuclei during phragmoplast expansion. Phragmoplast expansion begins symmetrically (T0 to T9.46) but switches to a polarized form (T24.0 to T59.21) after contact of the cell plate with one side of the cell cortex (arrowhead). Arrows indicate nucleus-associated microtubules that interact with the cell cortex. N, daughter nuclei, which have accumulated AtEB1a-GFP along their surfaces; T, time (min). Bar = 8.3 μm.
The final positioning of the cell plate also involved phragmoplast reorientation. Figure 8 (using an inverted grayscale image to visualize the cortex more clearly) shows an actively expanding phragmoplast rotate through 90° before selection of the division plane. Note that this rotation occurred in the same cell as illustrated in Figures 5A to 5D, underlining the fact that phragmoplast rotations (three cells) were only observed in cells undergoing mitosis in the absence of a PPB. In these cases, the phragmoplast reorientation resulted in division occurring along the shortest axis of the cell (Supplemental Movies 5 and 6 online show spindle and phragmoplast morphogenesis in the presence and absence of a PPB, respectively).
Phragmoplast Rotations.
Time-lapse confocal laser scanning microscopy images (inverted grayscale) showing an actively growing phragmoplast rotate through 90°. This data set was collected from the same cell shown in Figure 5A, which was used to demonstrate mitosis in the absence of a PPB. Arrows indicate nucleus-associated microtubules that interact with the cell cortex. N, nuclei; T, time (min). Bar = 8.3 μm.
Astral-Like Microtubules
As shown in Figure 8 (but better seen in Supplemental Movies 2 and 3 online), a general and notable feature of mobile daughter nuclei was a subset of microtubules that emanated from polar caps of AtEB1a-GFP around the nuclear surfaces distal to the phragmoplast (see arrows in Figures 4, 7, and 8). These microtubules interacted with the cell cortex (see Supplemental Movie 2 online). The appearance of these microtubules, which were reminiscent of the astral microtubules of animal cell division, was cell cycle regulated. They gradually increased in length and number as cells progressed from metaphase (where a few, short microtubules emanated from spindle poles) to anaphase and telophase, during which numerous microtubules were observed radiating from the outer surfaces of the reforming daughter nuclei (see Figure 4). Like astral microtubules (Tirnauer et al., 2004), the average growth rate of these nucleus-associated microtubules was faster than reported for interphase rates of cortical microtubule polymerization (6.12 ± 0.29 μm/min; 52 microtubules versus 3.53 ± 0.26 μm/min; Chan et al., 2003).
DISCUSSION
The Distribution of AtEB1a around the Cell Cycle
EB1 is a well-established microtubule binding protein that preferentially attaches to the growing plus end of microtubules in animals (Morrison et al., 1998; Mimori-Kiyosue et al., 2000) and fungi (Straube et al., 2003). This was also recently found to apply to cortical microtubules in Arabidopsis suspension cells, with AtEB1a labeling the fast-growing ends of cortical microtubules in a comet-like pattern (Chan et al., 2003; Mathur et al., 2003; van Damme et al., 2004). Previously, we used this labeling pattern to show that microtubules of the cortical array are of mixed polarity and that comets arise from multiple dispersed sites around the cell cortex (Chan et al., 2003).
In this study, the nuclear periphery becomes active as a site for microtubule outgrowth and is particularly obvious in those cells without a PPB where the microtubules form a radial array. This is consistent with the association of nucleating material with the perinuclear region at specific stages of the cell cycle, as reviewed by Vaughn and Harper (1998). During preprophase/prophase, comets were observed to grow away from the nuclear periphery toward the cell cortex. Comets were difficult to detect during metaphase, but during anaphase they could be seen to grow from the poles toward the equator of the spindle. During telophase, comets grew from the outward-facing side of the daughter nucleus toward the cell cortex or from the inward-facing side to the equator of the phragmoplast. In all cases, it can be deduced from the fact that plus end comets grow away from the nuclear periphery that the minus ends are located at the poles or the nuclear surface. This is consistent with hook decoration studies that place the minus ends at the spindle poles in mitosis (Euteneuer et al., 1982) and in the vicinity of the nuclear envelope during cytokinesis (Euteneuer and McIntosh, 1980)—in both cases with plus ends meeting at the midline.
It is intriguing that in all of these loci where it can be inferred that minus ends are directed, AtEB1a-GFP is also accumulated. Although microtubules arise during mitosis in the general region of the nucleus, the association is not tight, for we observe both circumferential and radial movements of the minus end foci at the nuclear periphery. Also, we observe nucleation to occur during anaphase when the nuclear envelope is not completely reformed. For these reasons, it does not seem that the minus end loci are exclusively associated with the nuclear envelope but may be associated, for example, with other components of the endomembrane system such as the endoplasmic reticulum that is known to aggregate at spindle poles upon breakdown of the nuclear envelope (Hepler, 1980; Zachariadis et al., 2001).
The minus end localization of AtEB1a-GFP is consistent with findings from diverse organisms, ranging from yeast to vertebrates, that plus end tracking EB1 also gathers at the minus ends of microtubules (Bu and Su, 2001; Morrison and Askham, 2001; Askham et al., 2002; Rehberg and Gräf, 2002; Rogers et al., 2002; Tirnauer et al., 2002, 2004; Straube et al., 2003; Louie et al., 2004; Piehl et al., 2004). It is also consistent with our previous study on Arabidopsis cells, which showed that AtEB1a-GFP labeled foci at the minus ends of cortical microtubules from which microtubules could be seen to grow or shrink (Chan et al., 2003). This initial targeting of AtEB1a-GFP to the minus end can be explained by the idea that it acts as a repository for the EB1 required during plus end growth (Rehberg and Gräf, 2002). The fact that EB1 accumulates at the poles in suspension cells appears to be related to the elevated levels of expression under the 35S promoter of Cauliflower mosaic virus and the transient-expression system where spindle fluorescence was 16.7 (mean of 35) times brighter than in Arabidopsis plants under the native promoter. By contrast, polar accumulations were only occasionally seen in Arabidopsis plants transformed under the 35S promoter (2.5, mean of 9) and never seen in plants transformed with the native promoter nor has it been observed in stable tobacco (Nicotiana tabacum) BY-2 lines expressing AtEB1a-GFP (van Damme et al., 2004). We observed that the minus end labeling was less prominent in those cells displaying rapid plus end dynamics with the comet-like pattern (see Supplemental Movie 3 online) and that the fluorescence intensity of perinuclear foci was reduced or disappeared after the emergence of comets. We therefore suggest that the higher levels of overexpression obtained in Arabidopsis suspension cells with the 35S promoter of Cauliflower mosaic virus allow for AtEB1a-GFP to persist at the minus ends at levels that are less readily depleted by microtubule outgrowth.
Whatever the explanation, the ability of AtEB1a-GFP to mark sites at the nuclear periphery has allowed the identification of a subpopulation of premitotic cells that divide without a PPB, allowing us to compare their behavior with cells that divide in a more conventional manner.
Alternative Mechanisms for Spindle Pole Morphogenesis
The time-lapse studies showed that two mechanisms for mitosis exist side by side in a suspension culture, much as has been reported for cultured and embryonic Drosophila melanogaster cells (Goshima and Vale, 2003). In Arabidopsis suspension cells where there is a PPB, perinuclear AtEB1a-GFP is sorted into two polar caps perpendicular to the PPB before NEBD. This is reminiscent of the redistribution of γ-tubulin into bipolar caps in preprophase onion (Allium cepa) root tip cells (Liu et al., 1993). These premitotic bipolar caps of AtEB1-GFP transform seamlessly to locate to the spindle poles after NEBD. In most of the cells, however, there is no PPB, AtEB1a-GFP is not prepolarized into caps, and the bipolar spindle is formed after NEBD. This suggests that in these cells the polar accumulations only become focused upon the assembly of the spindle in what would appear to be a microtubule-dependent process.
Recent reviews of spindle development have focused on an inside-out mechanism for plant cells and meiotic animal cells that possess no centrioles/centrosomes. In this mechanism, microtubules first associate with chromosomes before being self-organized into bipolar arrays by motor proteins (for review, see Compton, 2000). Using beads coated with bacterial DNA, incubated in centrosome-free extracts of Xenopus, Heald et al. (1996) showed that this method accounted for spindle development in the absence of centrosomes and kinetochores. In this mechanism, microtubules are nucleated in the vicinity of chromatin and then sorted by plus end–directed BIM C kinesins into two antiparallel sets with plus ends overlapping at the midline. Then, the distal minus ends are focused into poles by the action of minus end motors that could convey materials such as AtEB1a-GFP to the poles as spindle bipolarity develops.
In the alternative outside-in or search and capture mechanism (Compton, 2000), centrosomes are the dominant centers of microtubule nucleation and they determine the number and position of the spindle poles. Thus, spindle bipolarity in such cells relies on both the proper duplication and separation of the centrosomes. This situation would seem to be analogous to the process we observe in cells possessing a PPB, where AtEB1a-GFP labels two polar caps before NEBD; that is, before the microtubules and the chromatin mix. During the premitotic period in such cells, the three-dimensional arrangement of cytoplasmic strands that radiates from, and suspends, the nucleus is reorganized to form a two-dimensional division plane—the phragmosome with the PPB at its perimeter (Lloyd, 1991). The fact that the perinuclear AtEB1a-GFP does not separate into two polar caps in those cells without PPBs implies that this premitotic cytoplasmic reorganization may be necessary for bipolar cap formation. In this scenario, AtEB1a-GFP is moved to opposite poles of the premitotic nucleus as part of the process that draws most of the cytoplasmic strands into the PPB/phragmosomal plane and leaves polar strands to radiate from the nucleus perpendicular to the division plane. This premitotic polarization involves actin as well as microtubules, and destruction of the polar strands can cause the spindle axis to be misaligned (Lloyd and Traas, 1988). The formation of a tight PPB and the defining of the spindle poles (and, by extension, polar caps) in premitosis may also be connected and has been observed in other studies. For instance, Granger and Cyr (2001) showed that nuclei not overlapped by a PPB do not form distinct spindle poles (see Figure 9 of their study). Abnormal PPB and spindle pole formation has also been observed in Arabidopsis plants deficient in the kinesin, KatA, where the PPB did not narrow to the same degree as wild-type cells (Marcus et al., 2003). Consequently, microtubules formed symmetrically around the nucleus, without clearing to form distinct spindle poles.
Model Showing Pathways of Spindle and Phragmoplast Morphogenesis in Higher Plants.
Cells undergo alternative pathways of spindle pole development depending upon the presence or absence of a PPB. In its presence (follow arrow 1), spindle poles form during prophase before NEBD in a manner consistent with outside-in (Compton, 2000) spindle assembly. By contrast, in cells without a PPB (follow arrow 2), spindle poles form during prometaphase in a manner consistent with inside-out spindle assembly pathway. Phragmoplast morphogenesis occurs via a single pathway (follow arrow 3) from remnants of microtubules present in the spindle midzone region in somatic cells. The subsequent stages of phragmoplast expansion, however, are flexible and can occur either symmetrically (follow arrow 5) or are polarized after contact of the cell plate with one side of the cell cortex (follow arrows 6 and 7). Polarized cytokinesis can occur with either stationary nuclei (follow arrow 6) or with nuclei that migrate with the growing front of the phragmoplast (follow arrow 7). Nuclear migrations can also contribute to the outgrowth of the phragmoplast (follow arrow 6) and to the final position of the cell plate in the absence of a PPB, where the orientation of the spindle is indeterminate (arrow 4). Red dots indicate the cell cycle distribution of AtEB1a-GFP, which, apart from labeling at the midline of the phragmoplast, marks the location of microtubule minus ends.
In early models, the opposite outside-in mechanism was predicted from the existence of the prophase spindle (e.g., Wick and Duniec, 1983; Baskin and Cande, 1990; Lambert et al., 1991). However, the recent application of the inside-out mechanism to plants (Compton, 2000) overlooks the existence of polar caps and the prophase spindle. These observations underline the inadequacy of equating plant cells with meiotic animal cells because plant cell division can be brought about by both inside-out and outside-in mechanisms and that the latter can function in the absence of centrioles.
In a recent study by Goshima and Vale (2003), spindle pole development in somatic animal cells was similarly demonstrated to proceed via outside-in as well as inside-out mechanisms—the latter being activated when one of the poles lacks a centrosome, as in the centrosome ablation studies of Khodjakov et al. (2000). However, rather than thinking of the inside-out mechanism as a backup, Gadde and Heald (2004) have suggested that this model, which was once thought not to apply to somatic animal cells with centrosomes, may in fact operate in such cells, reflecting a natural redundancy. In plants, such a redundancy would be important for endosperm and meiotic cells that do not display PPBs and which, without the premitotic cytoplasmic reorganization that this implies, would divide by the inside-out route. It is unclear why this phenomenon is so prevalent in the Arabidopsis culture. This phenomenon is not seen in our tobacco BY-2 cell line that is grown alongside the Arabidopsis culture under similar conditions. The fact that the latter tends to divide in an undifferentiated mass instead of forming regular chains of cells like BY-2 does suggest that aspects of cytoplasmic polarization (of which PPB formation is an indicator) have been lost during culture.
Variable Phragmoplast Growth and Astral Microtubules
Variations in cytokinesis have been reported before (Venverloo and Libbenga, 1981), and Cutler and Ehrhardt (2002) used the term polarized cytokinesis to describe the situation observed in Arabidopsis plants where the phragmoplast contacts one wall first and subsequently expands as a crescent, not a ring. In their study, Venverloo and Libbenga (1981) showed that the nucleus remained anchored to the first-contacted wall throughout subsequent cytokinesis. Another form of polarized cytokinesis was observed in this study in which the daughter nuclei do not remain anchored relative to the first-contacted wall but move with the cell plate to the opposite side of the cell. By this means, the AtEB1a-decorated nuclei move with the leading edge of the phragmoplast. In this case, microtubules would seem to persist with the nucleus rather than redistributing away from it as shown in Figure 6. Because Cutler and Ehrhardt (2002) have shown that both central and polarized cytokinesis can occur a few cells apart in Arabidopsis roots, it is likely that the different mechanisms we observe similarly reflect a natural phenomenon rather than an abnormality of cultured cells.
Actively growing phragmoplasts, together with their nuclei, often rotated during cytokinesis before selection of the final division plane. This behavior was observed in cells that divided without a PPB, indicating that the PPB (when present) promotes the positional stability of these structures during cell division, as suggested by Gimenez-Abian et al. (1998). This could mean that the division site, which is somehow marked during preprophase, provides a reference point for cytoplasmic elements (e.g., cytoplasmic strands) to be used later during phragmoplast outgrowth.
An interesting feature of the phragmoplast reorientations observed in this study was the aster-like polar microtubules that extended from the upper surfaces of the daughter nuclei toward the cell cortex, moving around the circumference of the cell as the daughter nuclei and phragmoplast migrated. Astral microtubules are rarely mentioned in plants and have been suggested to be remnants of the nucleus-associated radial microtubules of the preceding interphase (for review, see Baskin and Cande, 1990). However, like the astral microtubules of animal cells (Morrison and Askham, 2001; Tirnauer et al., 2002), these nucleus-associated plant microtubules appear to be cell cycle regulated: their growth rate was almost twice as fast as interphase microtubules and they became more prominent from anaphase to telophase.
In animals and yeast, forces generated by growing astral microtubules are thought to push against the cortex; alternatively, cortically bound motor complexes reel in the astral microtubules (Kusch et al., 2002), thereby helping to orientate the spindle axis. In plants, these astral-like microtubules could assist with nuclear migration, helping to orientate the nuclei (and/or phragmoplast) relative to the plane of cell division and providing support for the growing edge of the phragmoplast as it moves across the vacuole. However, phragmoplast expansion probably involves more than microtubules, and there are indications that actin filaments may be involved in spindle rotation as well as in the centrifugal guidance and proper orientation of the new cell plate (Lloyd, 1991).
In conclusion, this study has shown that both spindle development and phragmoplast expansion can occur in different ways that may be related to the variable degrees of premitotic cytoplasmic reorganization according to the presence or absence of a PPB. Some of these behaviors, including the inside-out spindle development pathway and mobility of the daughter nuclei (phragmoplast reorientations), may therefore serve as alternative mechanisms to rescue cell division in the absence of premitotic cues that predict the plane of cell division (Figure 9). This would be consistent with the fact that sex cells and endosperm divide without a PPB and appear to erect a spindle by the inside-out mechanism (Smirnova and Bajer, 1992). Division is not confined to small meristematic cells but can occur adventitiously in large highly vacuolated cells under a variety of natural and experimental conditions. Alternative pathways provide the flexibility for plants to adapt to varying circumstances.
METHODS
Plant Material
Suspension cultures of Arabidopsis thaliana (Columbia) were maintained as described in Chan et al. (2003). Arabidopsis seedlings were grown on plates containing 3% (w/v) phytagel, 0.5% (w/v) sucrose, and 0.43% (w/v) MS powdered medium, with macro- and micro-elements (Duchefa, Haarlem, The Netherlands). Plates were incubated for 2 d at 4°C and then transferred to the growth room at 22°C under continuous lighting. Seedlings were used for confocal microscopy after 3 to 5 d.
Transient Tranformation of Arabidopsis Suspension Cultures
Cells were transformed with either AtEB1a-GFP or HDEL-YFP as described by Chan et al. (2003), except that 200 μL of the Agrobacterium tumefaciens suspension was used to transform cells and that suspension cultures were supplemented with 50 μg/mL of acetosyringone before transformation. Expression of AtEB1-GFP was visible within 2 d, after which cells were subcultured to encourage division. The HDEL-YFP construct was obtained from Chris Hawes (Oxford Brookes University, Oxford, UK).
Preparation of the Native Promoter:AtEB1-GFP Construct and Preparation of Stable Lines
The AtEB1a gene (introns and exons) plus 1 kb of sequence upstream of the open reading frame was amplified from genomic DNA and inserted into the pGEM-T-Easy vector (Promega, Southampton, UK) according to the manufacturer's instructions. The primers used for the PCR reaction were as follows: 5′-AAGGTACCTAACCGATTAGACTTGCAGG-3′ (AtEB1-3-kpn1) and 5′-TTGTCGACTTGGCTTGAGTCTTTTCTTC-3′, which amplified a product of 3263 bp. These primers also flanked the genomic DNA fragment with SalI and KpnI restriction enzyme sites. These sites were then used to cut the PCR product from the pGEM-T-Easy vector and ligate it into the pGR4 plasmid (provided by Gethin Roberts, John Innes Centre, Norwich, UK), which contains a C-terminal GFP tag. The native promoter:AtEB1a-GFP construct was checked by sequencing and transformed into Arabidopsis plants by transformation with Agrobacterium (strain gv3101) using the floral dip technique (Bechtold et al., 1993). To select for transformants, T1 generation seeds were harvested and sown on MS powdered medium containing 50 μg/mL of kanamycin. Resistant seedlings were transplanted to soil and the T2 seeds analyzed for the segregation of kanamycin resistance. The χ2 test was used to predict lines containing a single copy of the transgene. Homozygous lines were then identified in the T3 generation and used for microscopic analysis of GFP expression. The 35S:AtEB1a-GFP construct was transformed into plants as described above.
Immunofluorescence
Transformed cells were fixed for 1 h in PME buffer (50 mM piperazine-N,N′-bis-[2-ethanesulphonic acid], 5 mM magnesium sulphate, and 5 mM ethylene-glycol-bis[β-aminoethylether]N,N,N′-tetracetic acid) plus 0.025 M sorbitol (PMES), containing 4% (w/v) formaldehyde and 0.1% (v/v) glutaraldehyde. After three washes in PMES, the cells were stuck to poly-l-Lys–coated slides and extracted in PMES containing 0.5% (w/v) Onozuka R-10 cellulase (Yakult, Tokyo, Japan), 0.05% (w/v) Pectolyase Y-23 (Yakult), and 0.05% (v/v) Triton X-100. Cells were labeled with rat antitubulin (YOL 1/34; Harlan Sera-Lab, Loughborough, UK; 1:50) and rabbit anti-GFP (Molecular Probes, Eugene, OR; 1:100) diluted in PMES containing 1% (w/v) BSA. Primary antibody binding was then detected, after three washes in PMES, by applying anti-rabbit fluorescein isothiocyanate–conjugated (DAKO, Glostrop, Denmark; 1:200) and anti-rat CY3-conjugated antibodies (Amersham Biosciences, Little Chalfont, UK; 1:500) diluted as described above. The cells were washed four more times, stained with 4′,6-diamidino-2-phenylindole (DAPI; 3 μg/mL), and then mounted in Citifluor antifade mountant (City University Chemistry Department, London, UK).
DAPI Staining of Cells
DNA was visualized by adding 25 μL of Triton X-100 (10% [v/v] stock; 0.25% [v/v] final concentration) and 5 μL of DAPI (3 μg/mL stock; 0.015 μg/mL final concentration) to 1 mL of suspension cells.
Confocal Microscopy and Image Analysis
Images were acquired using a Bio-Rad MRC 1024 confocal microscope (Hercules, CA). Cells were maintained in a modified chamber fitted with a gas-permeable membrane (bioFOLIE; VivaScience, Goettingen, Germany). Images were taken at 5- to 600-s intervals. Both EB1-GFP and HDEL-YFP were excited at 488 nm and visualized using a 515-nm long-pass filter. Microtubule growth rates were measured in the time-lapse images using the line and reslice tools of ImageJ (http://reb.info.nih.gov/ij/).
Quantification of Fluorescence Levels in Spindles of Transgenic Plants and Suspension Cultures by Confocal Microscopy
For quantification experiments, single transgenic lines of AtEB1a-GFP under control of either 35S or native promoter were selected. In both cases, AtEB1a-GFP was uniformly expressed, and cells were phenotypically indistinguishable from those in control plants. Nonsaturated images of spindles labeled by AtEB1a-GFP were collected using a VisiTech spinning disc confocal microscope (Sunderland, UK) fitted with a ×60/1.4 oil objective lens. GFP was excited using the 488-nm line of an argon ion laser and the emitted light filtered through a 500- to 550-nm band-pass filter. Spindle fluorescence was detected using a Hamamatsu Orca ER cooled CCD camera (Hamamatsu City, Japan) with 4× binning, set at 1-s exposure time and 1-μm z-step intervals. Images were analyzed using ImageJ, z-sections were projected (max), and the average intensities of entire spindles calculated using the measure tool. Max z-projections ensured that the calculations were performed on complete spindles.
Sequence data from this article have been deposited with the EMBL/GenBank data libraries under accession numbers At3g47690 (AtEB1-a), At5g62500 (AtEB1-b), and At5g67270 (AtEB1-c).
Acknowledgments
We thank Gethin Roberts, Pascale Rossignol, Hannah Jones, and Nuno Geraldo for assistance in molecular biology and Guoie Mao for useful discussions. We also thank Chris Hawes for supplying the HDEL-YFP construct. G.C. and J.C. would like to dedicate this paper to the memory of Dagon Prince.
Footnotes
The authors responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) are: Jordi Chan (jordi.chan{at}bbsrc.ac.uk) and Clive Lloyd (clive.lloyd{at}bbsrc.ac.uk).
Article, publication date, and citation information can be found at www.plantcell.org/cgi/doi/10.1105/tpc.105.032615.
↵1 These authors contributed equally to this work.
↵w⃞ Online version contains Web-only data.
- Received March 10, 2005.
- Revised April 13, 2005.
- Accepted April 14, 2005.
- Published May 6, 2005.
References
